Home

SCIL_Image User Manual

1. int destination_function IMAGE image int n COMMON_LINE xcom line e image is the destination image e nis number of times to the function has been called when first called n 0 e com_line is a pointer to the COMMON_LINE structure used in the convert operation e The destination function is called after each call to the source function and must read the data from the common_line and put it in the image at the position specified by y and z If a 3D image is converted into a 2D image then the function may stop if the z value is more than zero or take another appropriate action In the default implementation for example the values in other z positions than 0 are compared with the values already stored in the 2D image and the maximum is taken as the value for that pixel This method provides a good 2D representation of most 3D images e The destination function must be able to read the data in the common_line regardless of whether it is in long word format or in double format The fields hint_min and hint_max can be useful for the conversion e The return value of the function should indicate if the function can handle more lines of data If more data can be processed a non zero value should be returned If for some reason the destination function cannot continue processing it should return a 0 zero to stop the conversion The source of the default grey value function is listed below for reference int g_
2. Pointer for future use do not use IMAGE The members of the IMAGE structure can be referenced through macros for convenience These macros also reside in the file image h They are ImageTypeldent ip ImageTypeSpec ip ImageWidth ip ImageHeight ip ImageDepth ip ImageSlice ip OldImageWidth ip OldImageHeight ip ip gt o_leny OldImageDepth ip ip gt o_lenz ip gt type_ident ImageName ip ip gt image_name ip gt type_spec ip gt lenx ip gt leny ip gt lenz ip gt slice ip gt o_lenx ImageDispName ip ip gt window_name ImageClut ip ip gt clut ImageGraphics ip ip gt graphics ImageROI ip ip gt roi ImagelIn ip ip gt in_descript ImageOut ip ip gt out_descript ImageFlags ip ip gt flags ImageOpCount ip ip gt op_cnt ImageInfo ip ip gt image_info ImageSize ip ImageWidth ip ImageHeight ip ImageDepth ip 9 6 Programming with Image Image 2 1 Library User manual The first two fields of any image type descriptor have a fixed meaning and are therefore present in a separate structure that can be referred to typedef struct type_fixed int type Type Identifier void data Pointer to start of Image Data IMAGE TYPE FIXED The macros to access the fixed members of any image type descriptor ImageInType ip IM
3. a i E Jello tif Trui ids Maan im Orka256 ics Orka256 ids E Gogh tif Si Sche Figure 3 4 The file browser The Open dialog box disappears and we return to the readfile dialog box click OK or Do It The command readfile is now executed and an image appears in window A Note that at the same time the text readfile c scilimage14 images schema dat A 300 300 appears in both the Worksheet and the History windows the latter may still be an icon This will always happen when you execute a command we will demonstrate the use of these windows in Session Two A g2D 256 256 Figure 3 5 The schema image 3 6 Getting started SCIL_Image 1 4 User Manual The image is displayed in the window labeled A This window is called a viewport to image A you may define several viewports for the same image which we will do later in this session When you use the scrollbars of the viewport you see that they move the viewport over the image scroll window A using the scrollbars This is useful when looking at parts of large images On the schema viewport a black border scrolls in to indicate where the actual image data ends You can scroll only until one of the corners of the image reaches the center of the viewport The image in A is a two dimensional grey valued image of 256 by 256 pixels picture elements This is indicated on the title bar by A g2D 256 256
4. File Format the types image clut v_object histogram In several examples you have seen the type image used This is actually not a new type but a different name for a type choice variable with some additional properties In the next session we describe these derived types and show how you can construct them yourself Very Advanced SCIL_Image New Types Whenever SCIL_Image encounters a type which is unknown it calls a list of functions that try to interpret the type You can add functions to this list This gives you the opportunity to define types of your own if you need something beyond the special types provided with SCIL_Image To add a type handling function you must use the standard function set_extra_type_func This function takes as its argument a pointer to a function int set_extra_type_func int fun_ptr The pointer is to a function e g my_type_func which contains the behavior of your new type The synopsis of this function is int my_type_func char name char type char deflt char minstr char maxstr char prompt Internally the first thing your function should do is to compare the input t ype to the name of the new type that it defines If the given type is not for your function you should return the value 0 If the type is for your function then you must change the given information into one of the know types and return the value 1 5 14 Advanced SCIL_Image SCIL_Im
5. You can inspect the individual pixels by pointing and clicking click on image A to select it click again on image A and drag In the status bar a set of numbers appears which change when you drag 156 112 214 The first two numbers indicate the position where you clicked the third number is the grey value For an image of type g2D this is an integer The position numbers are the coordinates of the pixel picture element of the stored image not of the point in the viewport For example the capacitor on the right always has coordinates 150 164 no matter how the viewport has been scrolled There are a number of image types in SCIL_Image including real valued images f complex valued images c binary images b color images col We will come across them in the course of the next demo session You can have several viewports connected to the same image Let us generate a second viewport on image A select Display create_display select Image to connect display to A click OK A new viewport appears also labeled A g2D 256 256 since it displays the same image To show that these viewports are indeed displaying the same image let us perform an operation on the image select Image Histogram equalize Getting started 3 7 SCIL_Image 1 4 User Manual select Output Image A click OK Image A is processed in place and the changing result is immediately reflected i
6. ep errbuf strlen errbuf if error message print it if ptr amp amp ptr sprintf ep Ss ptr ep errbuf strlen errbuf if recursive print recusion depth if imerr gt fstack rcount i 1 gt 1 sprintf ep rec_depth d n imerr gt fstack rcount i 1 else sprintf ep n ep errbuf strlen errbuf Creating an Image Application 13 5 Image 2 1 Library User manual return Function im_handle_error This function is called when an err by the image library In the main 1 stack x void im_handle_error IM_ERROR_STATUS i data int cl IM_ERROR_STATUS imerr imerr im_error my_dump_err_stack amp imerr errbuf image_output IMO_ERROR errbuf return This function handles the text outp void WINAPI my_output_func int stream switch stream case IMO_INSTRUCT printf My Image Inst printf text break case IMO_WARNING printf My Image Warn printf text break case IMO_ERROR printf My Image Erro fprintf stderr text break case IMO_OUTPUT default printf text break return This function is subscribed to or has occured and has been handled oop it is subscribed to the error m_error void dummy int mess int print the info to a buffer print the buffer r ut from the image_output function char text ruction
7. spb_unsubscribe data NULL handle_images break default break Creating an Image Application SAND 13 7 Image 2 1 Library User manual 8 Creating an Image Application
8. COMMON LINE com_line Our byte 2d version 12 12 New image types Image 2 1 Library User manual int byte_2d_conv_to_common IMAGE im int n COMMON_LINE com_line BYTE_2D_IMAGE g_im BYTE pix long data register long npix im_begin_func byte_2d_conv_to_common g_im BYTE_2D_IMAGE ImageIn im npix Byte2dImageWidth g_im ifen SSO First time called data long malloc size_t npix sizeof long if data return im_report_error byte_2d_conv_to_common IE_NOMEM no memory allocated for data set_common_line com_line COM_LONG data 0 0 0 0 1 0 0 255 0 data com_line gt data com_line gt x npix com_line gt y n com_line gt z 0 pix Byte2dImageData g_im pix com_line gt y npix while npix gt 0 datat pixt t if n Byte2dImageHeight g_im 1 im_end_func byte_2d_conv_to_common return 0 0 stop this is the last line im_end_func byte_2d_conv_to_common return 1 1 more lines are available The function conv_from_common should return OK 1 as long as it can store the data it is offered If it can not store the data it should return NOT_OK 0 which will stop the conversion Its header int type_conv_from_common IMAGE im int n COMMON LINE com_line For the byte 2d image type this function does t
9. Data representation of float images Floating point images consist of pixels each of which can have any real value within precision limitations Each pixel is represented by the C type float Though floating point images can contain more information than grey valued images they are far less common This is because computers used to work far more efficiently with integers and because float usually requires double the amount of space 32 bits per pixel Try readfile trui A eval B 0 005 A Image B has now become a float image as is confirmed in its title bar The image however is no longer visible due to the low values of the pixels By pressing the left mouse button in image B the value of the pixels can be examined XXX YYY VV VVV Note that the value which pops up for each pixel is a real number rather than an integer Usage of float images Floating point images are used in situations where accuracy is more important than speed in particular when there are many mathematical operations on an image Examples of operations on float images The trigonometric functions in SCIL_Image all operate on and result in float images The following function calculates the sine for each point in image B sin_im B C The result is almost invisible because all pixels have a low value In order to boost the contrast of the image all values can be multiplied with a certain factor To estimate that factor the maximum v
10. MENU menu menu options OPTIONAL If the command is to appear under one or several menus it must be registered with those menus This is accomplished with the keyword MENU followed by one or more menu names The menus must have been previously declared in the lt menu entry gt The signs are optional and only for compatibility with older versions of SCIL_Image The menu options keywords follow the menu names They are used to influence the layout of the menu and to change the behavior of the automatic generation of dialog boxes The menu options keywords are MENUSEP A separator line is drawn directly beneath the item in all the menus this item is present MENUKEY key key is a single character that activates the item in the current menu In the menu the first key character of the item name is underlined SHORTCUT keyl lt modifier gt Fnr key is a single character that combined with the shortcut key activates an item from anywhere in the menu tree lt modifier gt Fnr is a function key with a modifier key Shift Ctrl or Alt that activates an item from anywhere in the menu tree nr is the number of the function key The shortcut key is displayed next to item in the menu Since the shortcut keys activate an item anywhere in the menu tree they must be unique BOLD ITALIC UNDERLINE These keyword change the text style of the menu item These keyword may not be implemented on all platforms NOPRINT The command is not printed
11. The command does not execute and you get a warning box WARNING AN SCIL_Image Error Function gauss Message Function is not implemented for image type COMPLEX_2D Stop More info Figure 3 11 An alert box 3 16 Getting started SCIL_Image 1 4 User Manual You exit from this by clicking Continue You can convert the complex image to a float real valued image by taking its real part using the command found Arithmetic Complex_based select Arithmetic Complex_based real_im choose Input Image C choose Output Image D click OK The type of the new image in window D is f2D a floating point image The image values are not integers between 0 and 256 as in the case of g2D but real numbers you may check this by the click and drag method of viewing the image content pixel by pixel with display in the status bar Some of the numbers are negative and they are displayed as black Somewhere in the analysis of any image you will probably create abinary image This is a two valued image usually used to denote which pixels belong to objects and which to the background Let us take an example where we want to perform measurements on the objects in the image select Image I_O readfile click the Browse button find the file scilimage14 images cermet im double click on this file s name click OK To convert this into an appropriate binary
12. that line in the Worksheet or History window clicking and executing e You can also edit a command in the Worksheet not in the History window using the normal method mouse selection and re typing before you re execute it In the same way you can remove lines The above even applies to the RunEdit window mouse selection and Enter key executes your selection as if it were in the Direct Command Mode The Macro Mode After you have worked in the Direct Command Mode for a while you will notice that some command sequences have a tendency to re occur in your work You can of course type them in once and re execute them using the Worksheet and or History windows but SCIL_Image provides an easier way You can make a file that contains such a sequence of commands and execute the file as a whole Such a file of direct commands is called a macro To make a macro do the following 1 Go into the RunEditor using File New C Program under the menu system 2 Type in the desired sequence or copy it from the Worksheet or the History window using the mouse select and the commands in the Edit menu The C Interpreter 4 5 SCIL_Image 1 4 User Manual 3 Save the file using File Save in one of the directories listed in the environment variable MACRO see Chapter 2 for example in c scilimage14 demo macro As a filename use a name of your choice followed by the suffix mac for macro 4 Exit th
13. x x x 1p University of Amsterdam TNO Institute of Faculty of Mathematics Applied Physics Computer Science Physics and Astronomy Kruislaan 403 P O Box 155 1098 SJ Amsterdam 2600 AD Delft The Netherlands The Netherlands Stieltjesweg 1 2628 CK Delft The Netherlands SCIL_Image version 1 4 User Manual WINDOWS version May 1998 Copyright notice Copyright 1992 1998 by University of Amsterdam Faculty of Mathematics and Computer Science Amsterdam The Netherlands and TNO Institute of Applied Physics Delft The Netherlands All rights reserved No part of this publication may be reproduced transmitted transcribed stored in a retrieval system or translated into any language or computer language in any form or by any means electronic mechanical magnetic optical chemical manual or otherwise without the prior written permission of TNO Institute of Applied Physics Delft The Netherlands Disclaimer It is believed that the information in this publication is accurate as to the date of publication this information and the software package which is described are subject to change without notice Furthermore University of Amsterdam Faculty of Mathematics and Computer Science and TNO Institute of Applied Physics make no representations or warranties as to the accuracy or completeness of this publication nor as to the accuracy or completeness of the software package it describes All other warranties express o
14. All pixels in C have been set to zero Converting images into other types It is sometimes useful to modify the image type without losing the image data as occurs with set_im_type The command convert is supplied specifically for this purpose readfile trui A convert A B FLOAT 2D The grey valued image data in image A has been converted to floating point data in image in B convert is capable of changing any type of image data into any other type Apart from convert there are a number of commands that also convert the type of the data in the image to another type all of which can be found in the Image Conversion menu When a 3D image is converted to a 2D image each pixel in the 2D image will contain the maximum value of the pixels in the corresponding locations in all slices This results in a specific 2D interpretation of a 3D image readfile chromo3d A convert A B GREY 2D The Image 2 1 library in SCIL_Image 6 9 SCIL_Image 1 4 User Manual Filling images There are several ways to fill an image with data For instance a data file may be read from disk The most common way is to specify the image as the target for an operation It is often useful however to set the image contents specifically For instance set_int A 128 sets all pixels of image A to the value 128 many commands to fill an image can be found in the Image Generation menu each of which generates a test image For instance to fill image B
15. Help will pop up the Help page for the command This is the more direct way to get the above Help page lt command gt prompts you for the parameters of a command while offering the default values do not forget to put a space between the command name and the question mark copy_im Input Image lt AIBICID gt A Output Image lt Al BICID gt B lt pattern gt list all commands with names matching lt pattern gt The pattern may contain a wild card which can match any sequence of letters and or symbols 2s p FUNCTION S set_aio_disp Sfp silo_to_comp sleep start_comp strcasecmp strcmp strncasecmp strncmp The C Interpreter SCIL_Image 1 4 User Manual The lt pattern gt help option also works on interpreted functions which you may have loaded in macros and gives some basic information For the interpreted program simpprog c defined in The Programming Mode Interpreted C functions and UFOs on page 4 7 we have load simpprog c charcount FUNCTION S charcount Interpreted function Return type integer Errors Warnings and Diagnostics The SCIL_Image C interpreter will complain if the syntax of a command is incorrect The diagnostics are intended to be self explanatory For instance forgetting a semicolon will yield int i int i gt syntax error However SCIL_Image always tries to interpret your statement if it is unambiguous even if it is not synt
16. conv_to_common byte conv_from_common byte pix_value_str byte get_image_window_info byte part_image_display byte 2d_part_image_display Next the basic functions as described here together with the low level functions listed at the end of this chapter must be compiled and added to Image Low Levels The prototypes of all the functions are located in a separate include file byte_2dp h ifndef BYTE_2DP_H necessary to avoid multiple definitions define BYTE_2DP_H file inclusion include image h include byte_2d h include roi h prototypes of the basic functions if defined __STDC__ defined INTERPRETED BYTE_2D_IMAGE byte_2d_create_image char int int int int void byte_2d_destroy_image BYTE_2D_IMAGE int byte_2d_copy_part_image IMAGE IMAGE int int int int int int int inta ints int byte_2d_copy_masked_part IMAGE BOOL_MASK IMAGE int int int int Pat int ant tnt iat dnt char byte_2d_pix_value_str IMAGE int int int void byte_2d_get_image_window_info IMAGE char int byte_2d_conv_to_common IMAGE int COMMON_LINE int byte_2d_conv_from_common IMAGE int COMMON_LINE BYTE l_byte_read_part_image BYTE int int int int Dnt int int DneE SA S void l_byte_write_part_image BYTE BYTE int
17. im_end_func point_object return lp 5 When called the point_im function waits until a mouse event in an image occurred or a key on the keyboard was pressed Its prototype int point_im IMAGE ip int x int y IM EVENT but Analysis of Images and Objects AIO 10 7 Image 2 1 Library User manual When it returns on a key press of the key board it returns the character value of that key the values of ip x y and but are not valid then When a mouse event happens in an image the image the location of the pointer in image coordinates and the mouse status event are returned through the ip x y and but parameters The function MousePress tests if the event is a mouse down event its prototype int MousePress IM_EVENT but It returns which of the mouse buttons was pressed testing for a non zero value means that pressing any mouse button is accepted as valid input 10 8 Analysis of Images and Objects AIO Chapter 11 Bitmapped binary images In this chapter we present methods for fast morphological binary image processing using a bitmapped representation of binary images instead of representing binary images as bitplanes inserted in grey value images The bitmap data structure is a very efficient representation both in terms of memory requirements as in terms of algorithmic efficiency as the CPU operates on 32 pixels in parallel The algorithms described in this chapter are capable
18. image in which everything lighter than a given grey value is madeobject displayed as red and everything darker than that grey value is madebackground displayed as black select Image Conversion threshold click OK EF B b2D 256 256 OE x A Figure 3 7 A thresholded image A binary image now appears in window B marking all pixels that were less than 128 in greyvalue red you may check this by the point and drag method to display positions and values in the status bar We can use the zoom and pan tool to examine this in more detail select image B with the right mouse button A green border appears to show it has been selected Note that there is also still a green border around the selected viewport on image A Now when you zoom and pan the display in both viewports is affected play around with zoom and pan When you de select the viewport A from the zoom and pan tool by a right mouse click in the viewport viewport B is then displayed in the interaction window and the actions only affect that viewport On the de selected window the final display settings of zoom and pan Getting started 3 9 SCIL_Image 1 4 User Manual remain Thus you can use the zoom and pan tool to permanently adjust the display of your images You can close the zoom and pan tool completely by double clicking on its close box or by minimizing it Again the final display settings then remain valid on the selected images
19. is the numeric event ID data is a pointer to the optional message specific data The type of the publisher object defines the messages published and the data accompanying those messages This means that different types of data can accompany the same message when published by different object types Processing messages How to act when receiving a message is up to the subscriber The publish and subscribe mechanism does not impose any obligatory actions to certain messages except for the SPB_DESTROY message as explained in Messages on page 8 5 However when processing messages please keep the following in mind e When a publishing object has more than one subscriber it is undetermined in which order these functions are called The actual order may even change in future versions without notice e The current implementation of the publishing mechanism has a depth first strategy This means that new messages get priority E g Object B and object C are subscribed to object A and A publishes a message Object B receives this messages and decides to publish an other message itself This new message is first sent to the subscribers of object B before the message of object A is sent to object C 8 4 Publish and Subscribe Image 2 1 Library User manual Messages In the include file spublish h we give several often occurring events pre defined message IDs Also a number of structures are defined for passing mess
20. left II left X right _LOWEST Table 6 3 operater priority in eval The Image 2 1 library in SCIL_Image 6 19 SCIL_Image 1 4 User Manual The amp is a maximum operator and the amp is a minimum operator Operators on the same line in the table have equal priority Left associativity means that if multiple operators of equal priority are used without brackets then the leftmost has the highest priority The operators and can be used for neighborhood operations which means that not only one pixel can be referred to but also the pixels surrounding that pixel For instance eval B A A 1 0 A 0 1 A 1 0 A 0 1 Adds the pixel and its four neighbors for each pixel in the image and stores the result in the correct location in image B In this expression A 1 0 indicates the left hand neighbor of the pixel and A 0 1 is the pixel below The first number is the x distance from the pixel and the second number is the y distance When using offsets like this in an expression problems occur at the borders of the image because the pixels on the edge do not have a neighbor on one or two sides For this situation the second parameter of eval border value is used If border value equals 1 then the edge will work like a mirror when a pixel outside the image is pointed to a pixel within the image will be used The following example should make this clear eval D A 256 0 1 It sh
21. lt parent menu_name gt comment lt menu name gt A menu entry starts with a dollar sign followed by the name of the new menu lt parent menu gt Optionally the new menu name is followed by a list of parent menus A parent menu is referenced by a dollar sign followed by the name of the parent menu The new menu will be added as a child sub menu to each of the parent menus All parent menus referred to in the description must have been specified before comment Any text after the list of parents will be considered as a comment for your own use and will not be interpreted by the system 5 4 Advanced SCIL_Image SCIL_Image 1 4 User Manual translate entry the lt translate entry gt contains C type definitions for the CDF types used in function argument specifications For each of the argument types of commands described in the CDF files SCIL_Image needs to know the exact C type to be able to call the function correctly The syntax for the type definition is lt translateentry gt TRANSLATE cdf type C type TRANLATE cdf type C type A translate entry starts with the key word TRANSLATE The cdf type is an argument type that is not also a C type used to describe the types of the command parameters in the ARGS section of a lt command entry gt A complete listing can be found in command entry on page 5 6 The C type is the basic C type of the corresponding cdf type E g an odd
22. macro i v lt file gt execute macro file rmvar remove old variables and free allocated space run args interpret loaded program including main time lt command gt time a command lt pause break key gt interrupt program execution lt pattern gt list functions with names matching lt pattern gt lt command gt prompt for parameters of command lt Ctrl Enter gt pop up dialog box of selected command lt Ctrl H gt pop up Help facility The C Interpreter 4 9 SCIL_Image 1 4 User Manual chain lt filename gt args The chain command loads the file and starts running it immediately So instead of load example c run you can just type chain example c Please note the difference between the chain command and the macro command chain loads file with a fully correct C syntax a C program macro loads file that contains direct SCIL_Image commands a macro The chain command is found in the SCIL menu as SCIL chain list start end The text of the loaded program can be displayed with list rmvar include image h list 5 Displays line 5 of program text list 5 10 Lists lines from 5 to 10 list 10 Lists from start to line 10 list 10 Lists from line 10 to the end load lt filename gt The load command loads program text into SCIL_Image The size of a program is only limited by the amount of RAM available When no filename is entered SCIL_Image will prompt for one The following
23. problems To obtain optimal execution times special implementations have been developed for the common isotopic erosion and dilation Binary image processing functions are discussed in the chapter Bitmapped binary images 7 6 Introduction to Image 2 1 Image 2 1 Library User manual Appendixes The appendices of the manual contain additional information on Image Appendix ICS file format description The ICS format used to store images on disk is a proposed standard for the storage of image data This format is described in detail in an article supplied in this appendix Introduction to Image 2 1 7 7 Chapter 8 Publish and Subscribe This chapter is a description of the publish and subscribe message passing mechanism and its use in the Image library Image 2 1 Library User manual General aspects Publish and subscribe is a generic message passing mechanism in which an object broadcasts messages about important events and changes regarding itself Other independent objects can receive these messages as a means of getting information about changes to the publisher The publishing objects do not know if and how many objects will actually get these messages and how they react upon them Objects that want to receive messages from a publisher must subscribe to that particular object and messages will be sent to them only This way handling events generates little overhead The Image library uses th
24. 1 image types available in Image Writing your own image processing routines Image provides a flexible object oriented image processing infrastructure Image types are divided into class and sub class of objects Function overloading supports class specific operations Through the Image infrastructure a number of general purpose services are provided which support image type specific operation development The type and dimensions of input images are checked Output images are automatically converted to the correct type and dimensions based on the operation and input image s Furthermore the infrastructure supports the processing of arbitrary shaped region of interest ROI Introduction to Image 2 1 7 5 Image 2 1 Library User manual The chapter Programming with Image contains a detailed description of the infrastructure in the Image library It also explains how to write new image processing functions which make use of the image data structures of Image Custom Image types Another strong point of the Image library is the ability of adding new image types without any changes to the infrastructure By strict separation of the image types and the infrastructure not having any knowledge of the details of the image types a new image type can be added without changes to infrastructure The implementation of a new type is achieved by supplying a handful of pre defined functions to the infrastructure The chapter New image typ
25. 1 Library User manual possible conversion routines have been written to transfer the data from a var_object to an image and vice versa A routine is also provided to convert the data in a var_object to another type The routines are var_object_convert changes the data into another type var_object_to_image transfers the data to an image image_to_var_object stores data from an image in a var_object A var_object is converted in the same way as an image A common_line is used to store one line of data which is then read by the receiving object see Data conversion convert on page 9 32 In fact the destination functions that are used to transport the data between the common_line and an image are the same functions that are used for image conversion This means that if the source or destination function for an image type is overruled by the user the conversion of var_objects to or from images is also overruled The source and destination functions which handle the transport between var_objects and the common_line however cannot be overruled The data contained in an image regardless of its size or type always fits in a var_object without loss of information Histogram objects The histogram data of images in Image are stored in histogram objects These histogram objects can be of any size and have up to 5 dimensions For each dimension of the histogram the center pixel value of the first and the last cell are stored as
26. 10 3 Labeling Objects sina an e a aia ai e tinea 10 3 Measuring individual objects seeeeeseseseeeesseseresressessrssressessrerressersreseeeseeseeseresee 10 3 Object manipulati n area ia a a area aar ao iper ie US cad esas cae 10 4 Image SOn ieee a dar de ee eE E a A E dedi eee R dp ean eae 10 4 Direct train pull Aeron 3 2650 iena n a aE A E e E Ea Gee RNT aS 10 4 An ALO sample Session an ioina R a E ei lateaadd 10 4 Interactive Me AS ULE mentanina aae a e Mees a A a iaoa eieaa ii 10 6 Implementation of the interaction part seeeseseeeeesesssssressessresrersersrerereseeseeseeesee 10 7 The point_object function ssssseeeseseesseseesseseresressrserssresseserssresseseresressesee 10 7 Chapter 11 Bitmapped binary iMages essseosssesssosssossseosssessosssoosssessse 11 1 Erosions Dilations and Logarithmic Decomposition eesseseeeseesessrsereerersrrsrerressesee 11 2 Algorithmic Iniplemeritati On i iz syed i satoes eas ececasteetin edad aotied ieee aaa steaks 11 4 Data Representa LOU cs 9 3 pores a asa E E tobe cadeetencecsncouees 11 4 Implementation of the Pixelwise Logical Operations cecceeeeeceeeeeeeteeeeees 11 6 Implementation of the Morphological OperationS ee eeeeeseceseeeeeeereeeneeens 11 7 Eval ation sinais a Aes teats dactnges E R ial tides a ook e a gat ata add 11 8 Pixelwise Logical Operations yic 4ic cae ee uel aut a aE 11 9 Morphological Transforms 2 22045 GAw ein Gh Mew Bi a Gas aes 11 1
27. 2 An index into the SCIL_Image manual Why SCIL_Image Interactive environments have always played an important role in image processing because of its visual nature Generally there are no off the shelf solutions available for image processing problems The visual evaluation of image processing techniques helps to find solutions to these problems The immediate feedback interactive systems provide reduces the time necessary to develop new applications Furthermore the ability to see the results of commands and to modify code or parameters within seconds brings new insights for example as to how sensitive an algorithm is to a small change in the image The natural alliance between interactive systems and image processing has led to the development of an enormous variety of software systems for image processing most of which are bound to specific hardware or tailored to a specific image processing application domain 1 8 Outlines of SCIL_Image SCIL_Image 1 4 User Manual However Preston concluded in his 1981 survey covering 72 image processing languages that the diversity in image processing environments is not justified since image processing applications are not so disparate as to require unique and specialized data types or different programming constructions Rather than creating a new highly specialized language it is better to make use of an existing programming language The functionality of existing programming enviro
28. 3 Getting started This chapter contains four sample sessions to get you acquainted with SCIL_Image Read this chapter if e You have never used SCIL_Image e You want to follow some typical action scenarios e You want an overview of the menu contents e You want to know how different interactive tools can be combined Do not read this chapter if e You are familiar with the SCIL_Image menu system and command interpreter e SCIL_Image has not been properly installed see Chapter 2 SCIL_Image 1 4 User Manual Before you begin You should already be familiar with basic Windows concepts such as e Using icons e Using the mouse actions point drag click and double click e Pulling down menus and choosing commands e Scrolling in windows e Resizing and moving windows e Terms such as dialog box list box and button If you are not familiar with these consult your Windows manual The Five Modes of Interaction with SCIL_Image There are five ways in which you can engage in image processing with SCIL_Image They are in order of complexity e viewing images and using interactive tools e using mouse and menu e command line typing and use of macros e making your own interpreted programs and reaching them from menus e making your own compiled version of SCIL_Image 1 SCIL_Image was originally written for systems using the UNIX operating system As a consequence it has a few properties that are not qui
29. CAL eg 05 Jee Gat lash E Gated Roussanne te Secession E 9 29 Oyerload Tables 25 osceatt ied cca e ccd sa cctrncicas gs aca en cite ds saatoun EEOAE EES 9 30 Overruling the default implementation gs sececees 2G ceeteveieie die edcdscega testes 9 31 Data CONVETSION CONVELT a a A E a i aa 9 32 Super type of an image line seeeseseeseeseseseseessersresrrssttsresrrestesseseresresseseresressesee 9 32 COMMON LINE Structures Sarsan aaae EE E AENT ES 9 32 SOUTCE IMNCUOM Specification din e e EER R EE S 9 33 Destination function specification sseeesssesseessessseessetessttssresserssersseeeesseesseese 9 35 User Specified Conversion siise ii ete Bees a i a 9 37 Var Object enea ea a a a aa aa a 9 37 Var bject Stuctire sona a a E eel A A 9 37 Programming with var_objects eeseseseseesessesesesressessrssressessresresseeseerreseesreseresee 9 39 ASE CS OE var objects se act en aE E E E cds table E E EERS 9 39 Conversion of var_objects to images and vice versa sssssssssesssessssseesseessre 9 39 Histograni StrUCt TE oi Bete ae E A E EAE ie ind E E 9 40 Programming with histogram objects essesesessesessesssssressessrerrerseesresereseeseeseresee 9 42 Chapter 10 Analysis of Images and Objects AIO e ssessssessoosseosssess 10 1 General Concepts in Microscopical Image Analysis sseseeeeseseeeeesrreresrresresrrseresee 10 2 Components of the AIO framework ssssssseseseesessresseseresressrsrrssressessresrenseseresressesee
30. Morphological Image Processing and Analysis Optical Engineering 26 7 623 632 July 1987 VERWER B J H Verwer L J VanVliet A Contour Processing Method for Fast Binary Neighborhood Operations Pattern Recognition Letters 7 27 36 1988 GROEN F C A Groen N J Foster A Fast Algorithm for Cellular Logic Operations on Sequential Machines Pattern Recognition Letters 1988 SERRA J Serra Image Analysis and Mathematical Morphology Academic Press 1982 Bitmapped binary images 11 11 Chapter 12 New image types This chapter describes how a user can create subtypes of existing image types and how new image types can be implemented in Image as if they were standard types Read this chapter e If you want to implement a new image type e If you want to know more about the image infrastructure Do not read this chapter e If you are not an experienced C programmer e Ifyou have not read Programming with Image Image 2 1 Library User manual Introduction The infrastructure of Image supports a strong separation of the image types for reasons of type safety and extensibility To enable the definition of new image types without having to change the infrastructure requires that no knowledge of image type details are incorporated in the infrastructure This means that all image type specific actions must be performed by functions tailored for the image type Calling the correct function for each image
31. SCIL The location of comfile At delivery set to c scilimage14 2 6 Setting up SCIL_Image SCIL_Image 1 4 User Manual SCIL_HELP The file containing the on line Help facility When you start to process your own images with SCIL_Image you can either store those in the directory c scilimage14 images or in a directory of your own In the latter case you have to change the environment variable PICT by adding on the names of the directories which contain your images When you write programs or generate macros you can keep those in one of the directories listed in the scilinit file or in one of your own directories which you must add to the environment variable MACRO SCIL_Image finds other initialization information in the scilinit file for instance on the number of image windows it pops up when starting Such information can be changed to your liking except that you cannot ask for more than four 4 default images Also you should not change the order of the initialization statements The specific settings of the size and position of your SCIL_Image window and of its subwindows such as the Worksheet and the History windows are stored in a file scil ini which SCIL_Image creates in the Windows directory It is updated at the end of every session with your most recent preferred settings When you restart SCIL_Image it will look the same as when you last left it Setting up SCIL_Image 2 7 Chapter
32. SIZES sie er ei i tee iE E E sats EE EES 6 9 Changing image typ eSe carinei ac a E E E TE 6 9 Converting images into other types essesseseeeseseesseseresresseserssressereresressessees 6 9 Filling magisa ascia a ody paz cera A e teas Aa tical a eT a ae 6 10 Region of interest ROI processing sesssssessesssessseeeseetesstessressresseesseressseesseese 6 10 Imas Ypes ene a a a a a a eineke 6 11 Grey valued images GREY_2D amp GREY_3D sssssesssssssessseresseeessressrrssessee 6 11 Data representation of grey valued images s ssssssessssseesseesseessessseessseessee 6 11 Usage of grey valued images esis y5 a Sess a ee en 6 12 Examples of grey valued Operatlons seid ciasegvatsevveceeegancdseaeedesncns cecgeesecesnases 6 12 Binary bitmapped images BINARY_2D amp BINARY_3D ee eeeeeeeeeeeeees 6 12 Data representation of binary images 4 siiiscecs sa ceadsstecsa ss cheese cedahend dander se tecsads 6 12 Usage Gf Vinary TAGES cis Fey anos cu a scot a seceus ecu geo acadh cus ANN a 6 13 Examples Of binary Operations 6 2 5 Aiea cele eisai Bw Ba aeons 6 13 Floating point images FLOAT_2D amp FLOAT_3D eee eeeeeeeseeeeeteeeeees 6 14 Data representation Of float images lt ds sdcsascesndeassntecassas pduas Setaezed sosneeeeaeteteaes 6 14 WISASE OL flot IMAGES ss neie seats ced dessy couse teu aus E A A E des 6 14 Examples of operations on float images cceesseceesceceeeeeceeeeeceeeeeeeteeeenaes 6 14
33. a var_object readfile chromo3d A pix_maxval A objl Axis Z show_var_object_info obj1 The information printed in the command window states that the type of data is PIXEL the same type as the pixels of a grey valued image Note that the var_object is 2 dimensional This data can be put in to an image with var_object_to_image objl B GREY_2D In image B the result of the operation a good 2D representation of the 3D image is visible The reason why the result of pix_maxval is not stored in an image directly is that the result can also be 1 dimensional or even 0 dimensional a scalar depending upon the dimensions of the input image and the parameters of pix_maxval The Image 2 1 library in SCIL_Image 6 25 SCIL_Image 1 4 User Manual Example 2 In the above example the result of taking the maximum of a 3D image was a 2D array of pixels made visible by putting the data in an image The result of the same operation on a 2D image is a set of 1 dimensional data This data can be used as input in other operations readfile orka256 A pix_maxval A obj2 Axis X lookup A B obj2 Example 3 The result of measurements in certain images can also be stored in a var_object readfile cermet A threshold A B invert_im B B label B C shape C obj1 0 0 obj2 The objects in image C have been measured The number of each object is stored in var_object obj1 and the area of that object is stored in the corresponding element of obj2 These v
34. ar i a aoe 7 5 Writing your own image processing routines ssssssessseeesssressresseesseeeseeesseee 7 5 C stom SIM ASS types nearne r a sae A I 7 6 Measurement of objects AIO caysgo cis seciic ca cade sucess sedans vedetatas a oecasanzeinens tates 7 6 ES VEDA AAT AS Ser n a ala Gasp eae dos Pataca eae eens wos 7 6 Appendixes eee ee e e e teat ae gcateosaecnte E N S T 7 7 Appendix ICS file format description ceeeecssececssccecesccesssccecssceesssceeesees 7 7 Chapter 8 Publish and Subscribe ccccsssssscsccsssssssccsssssssssscssssssees 8 1 Genet al aS pects siz koien ene ee Stes adel cases thet nl scree tel one thet on Stee cae tana 8 2 Object reguirementSs icc eda a i sta es Su ce a n a aias 8 2 Subscribing and unsubscribing to objects s sesesseeesseesseessessseeesseeesseessresseessee 8 3 Receive messages eeen a a a a ae A O E a N T TEET 8 3 Publishing Messages eriin n tended E ERE A at Jesh aN 8 4 Processing Messages ron a n a n a e a a a a i 8 4 Messages getn aera T E R E Balada ua e a tees E ae E 8 5 Publish and Subscribe in the Image library ssnesesssesesesessseesssessessseeesseessseesseesseessee 8 5 Top level publishes nis e a a aati ee aaa 8 5 linatge oo S e COE AE E E T 8 6 Color lookup table publishes sseeeseseeesseseesesssesressessrssressessresressersrerereseeseeseresee 8 7 Top level Histogram publishes s sseseseeeeseeessenssessseessseessseessressessessseressseessrese 8 8 His
35. as was done on the other lines 5 Save the changed comfile by clicking the save button on the button bar or choosing File Save 6 To create the new scilimage14 comfile G Workspace scilimag right click the comfile project and EE comfile files choose Build EP comfiles l files grab32 files J E grabcol files E24 overload files q Scilimagel4 files 7 In the output window a list of files must now be printed if not the cdf file you added was older than the comfile and must be saved again 8 Start SCIL_Image again either by hitting lt CTRL gt F5 in the devstudio or by double clicking its program icon as you did before As you see the MyMenu button is now present in the SCIL_Image menu bar The command myfunc is now an item of MyMenw as well as an item of the Options menu Let s try it select MyMenu myfunc A dialog box pops up for your personal command which you may execute by clicking OK The function my func is not yet a compiled function it is still loaded from the file myfunc c and theninterpreted It is possible to add the function to the library of compiled functions of SCIL_Image and so make it into a full fledged command but explanation of that will have to wait till Chapter 5 Getting started 3 29 SCIL_Image 1 4 User Manual This concludes the fourth sample session Making a New Compiled Version of SCIL_Image In this sa
36. by SCIL_Image but most JPEG files will be read SCIL_Image requires that JPEG files have the jpg extension The TCL format The TCL Image format is supported for historical reasons and can only be used for grey valued 2 D images One file is used per image containing both the image data and the header information about the dimensions of the image Comment can also be stored in this file One disadvantage of this format is that not all image types under SCIL_Image can be stored in this file format Another is that the header information is not in ASCII and therefore cannot be read or modified easily Although the format does not have an obligatory extension after the filename SCIL_Image only accepts files that have the dat extension which is therefore the recommended extension for this format We insist on this extension because we also support other formats with other extensions The AIM format The AIM format is also supported for historical reasons we only support reading files in this format it cannot be written This format consists of two files per image a header file and a data file The header is not in ASCII so it cannot be easily read or modified The header file of this format has the extension hd and the datafile the extension im If the header file is not present the dimensions 256 256 are assumed The Image 2 1 library in SCIL_Image 6 23 SCIL_Image 1 4 User Manual Non image data var_o
37. by Step In this section we assume that you have just received your SCIL_Image package That package consists of A CD ROM with the SCIL_Image package A Hardware Protection Key dongle A License Registration Form containing the license code string You need the following to run SCIL_Image 2 2 A 486 DX computer 33 MHz or better At least 12 Mb of memory A 1024x768 monitor with 256 colors recommended A hard disk with minimally 15 Mb of free space recommended Windows 95 or Windows NT 4 0 Microsoft Visual C compiler only if you wish to extend SCIL_Image s functionality Setting up SCIL_Image SCIL_Image 1 4 User Manual The installation procedure is as follows 1 Insert the CD ROM in your CD ROM drive and start the installer Getup exe by double clicking its icon 5 2 The installer the workings of which should be self explanatory will install SCIL_Image in a directory of your choice When finished thesetup program will have created a program group and a program item which is represented by the SCIL_Image icon G SCIL_Image 3 Before you can run SCIL_Image you will need to put the license code from the License Registration Form into the file license scl in the scilimage14 directory Use a standard editor to type in the license code carefully copying it including the difference between upper and lower case symbols 4 Non demo licenses Plug the hardware protection ke
38. command rmvar The program and all variables connected with it have now been deleted You can reload the program by typing load myfunc c To inspect the text of the loaded program list Several windows of text scroll by of the include file image h the first line in your program file and the Worksheet ends up looking like this Getting started 3 27 SCIL_Image 1 4 User Manual Et Worksheet Cig x 357 endif IMAGE_H File myfunc c 1 include image h 3 myfunc filename image level 4 char filename 5 IMAGE image 6 int level 7 8 readfile filename image 0 0 9 threshold image image level 10 contour image image 0 8 0 11 Figure 3 16 Listing code in the worksheet A full description of these and other commands for program development can be found in Chapter 4 The second part of this sample session shows how you can make your programs available as commands in a menu and make a menu of your own The information used by SCIL_Image to generate menus and dialog boxes resides in a file called comfile This file is generated from smaller files These are recognizable by a cdf suffix and all are stored in the scilimage14 prog comfiles directory You can add functions and menus of your own by editing the file myfunc cdf residing in scilimage14 prog example and following a procedure we will now describe by an example Let us assu
39. command loads the file demo c load demo meas c Filenames are arbitrary but it is advisable to use some sort of convention here we always use the suffix c to indicate files containing commands When loading a specific file the contents of the previously loaded file are overwritten and the global variables are removed Directly after loading a file the functions in the file are known to the C interpreter They can then be 4 10 The C Interpreter SCIL_Image 1 4 User Manual used in the Direct Command Mode provided that they do not use any of the global variables in the file since these will only be declared when the entire program is run logon lt filename gt logoff The logon command followed by a filename creates a file in which all direct commands given during a session are stored Later this file can be used as a macro In the following example the file last_session is opened and subsequent commands are stored in it The logoff command is used to discontinue command logging and close the file logon last_session printf Good night Irene n Good night Irene logoff macro last_session Good night Irene If a logfile is active and another logon command is issued the active logfile is closed and subsequent commands will be stored in the file specified in the most recent logon Why do you need logon logoff if you already have a History window If you always give innocuous commands there is n
40. eacaeerieiaanes 1 11 SCIL Menu and dialog Seneratoniesaiss oisaniecieecs 235 con Ledetadon nner sated semnantenteteees 1 12 AS Image MAT ISS 8 e ty scat a RER A cadseea ac usedeveug nagetasd O a is 1 12 mase TIN TEAS CUCL Sian i Seating etd tas EEEE R ca eee aoe 1 12 Binary mathematical Morphology eeeeceeeesecesececeeeeeceeeeeceeeeecseeeeesteeeesaes 1 13 Numerical analysis of objects in 1Mages ee eee see ceeeeeseeeeneeeeseenseeeseeeenees 1 13 Chapter 2 Setting up SCIL_Image csssssssscccsssssssccsscsssssssccsssssees 2 1 BUNS Up Step By Step eset saa ioen e E T nek ete 2 2 The SC nia oS Od T e a a a a E NS 2 4 Your SCIL Image Environment s 2 0 554 boiler ees a i iS 2 6 Chapter 3 Getting started ssssoceccssoocccessssooccssssscccecssosoececcssooeesessssosese 3 1 Before yOu De 10 ies sissies ssccivicsesichssaasa nnana n n E Gee R ev E EE 3 2 The Five Modes of Interaction with SCIL_Image eee eeceeeeeeeeeeeeeeeneeeeeteeeenaes 3 2 Session One Viewing Images ci vicccceisicacsass cweasicevsaccadeaveedisdgvsenavesacea soaseeensoacaaeisogeneeed 3 3 Session Two Using the SCIL_Image Menu System eeeseeeeeeeesseeeeeneeeeeeees 3 13 Session Three Using the SCIL_Image Command Line Mode eee eeeeeeeteees 3 21 Session Four Programming in SCIL_Image cee ceecceceeeeecseeeeceeeeeceeeeeesseeeeeaeees 3 26 Making a New Compiled Version of SCIL_Image 0 0 0 ee eeeeeeeeseeeeneeeeaeenteeeees 3 30 The C
41. expander the line editor and various SCIL commands Chapter 5 describes advanced use of the C interpreter including how to add new routines to the system and how to modify the user interface This is particularly useful for application developers 1 6 Outlines of SCIL_Image SCIL_Image 1 4 User Manual Chapter 6 deals with the image processing of Image as part of the SCIL_Image environment The philosophy and general behavior of the image processing infrastructure and user interface are discussed Part III is the programmers manual of the Image 2 1 library Chapter 7 Chapter 8 Chapter 9 Chapter 10 Chapter 11 Chapter 12 Chapter 13 is an general introduction to the Image 2 1 library and to the manual deals with the publish and subscribe message passing mechanism Image 2 1 uses this to communicate with any interface focuses on advanced use of Image particularly useful for the program developer of image processing functions describes the AIO library handles the philosophy and use of the binary bitmapped routines describes how to implement new image types is about writing an application with the Image 2 1 library Outlines of SCIL_Image 1 7 SCIL_Image 1 4 User Manual Command SCIL menu and Description dialogue system Eile Chapter 5 SCIL command expander Chapter 4 SCIL C Interpreter Image User defined E i ED imported Library Chapters 7 to 13 libraries Chapter 5 Figure 1
42. fragments can be entered in the interpreter literally When using Image without SCIL these fragments will have to be part of a complete C program AIO is not meant as a complete measurement system It simply provides useful primitives which can be used in C programs It should also be noted that more complete documentation of the individual functions can be found in the on line manuals Interactive measurement The point and click measurement of objects in AIO is of course only possible when a user interface is present In the SCIL_Image package a routine is supplied that returns the object that the mouse is pointing at the point_object function The following example shows how the results of the optical density measurement can be obtained by pointing with the mouse at objects in image C It is assumed the first part of the previous sample session at least up to the List__label command has been executed while object point_object c o_list printf sum f mean f stdev f n od_sum object od_mean object od_stdev object 10 6 Analysis of Images and Objects AIO Image 2 1 Library User manual Implementation of the interaction part The Image Library is designed to be independent of a user interface as a consequence the point_object is not a default part of the AIO package It is however a standard function of the SCIL_Image program To allow for interactive measurement in other user interfac
43. get_image_window_info put info on image in a string for use in a title bar part_image_display display a rectangular part of the image 12 2 New image types Image 2 1 Library User manual In this section we will explain the process of adding a new image type by implementing the byte 2d image type a two dimensional image with 8 bit pixels The basic functions for the standard image types GREY_2D BINARY_2D and FLOAT_2D are made available in the src_exmp directory in the standard distribution These functions serve as a Starting point to develop type specific service functions In the basic c and basiclow c files in these directories the interface functions for these image types are gathered It is important to understand that the interface functions are accessed through the overloading mechanism Therefore sometimes some extra parameters are in the function argument list Please note that the last three functions are only required when implementing a new image type in the complete SCIL_Image environment These three functions are needed by the display interface of SCIL_Image the Image library does not need them to support the image types in stand alone applications Defines and structures In order for the infrastructure to recognize the new image type atype_ident anda type_spec must be defined The t ype_spec is a number that is unique for the data representation of the image it must be an unique bit in a 32 bit w
44. image Figure 11 2 shows the linear array depicted in a two dimensional layout The numbers in the words are the indices in the linear array The border words depicted at the right of the image also serve as 32 pixel border at the left of the actual image word 17 is the right neighbor of word 16 and the left neighbor of word 18 The word with index 1 is needed to store the north west neighboring pixel of the left most pixel in word 9 rial l T a a es Eal feu 7 oe ieee ae i left bit right bit Figure 11 2 Bitmapped Image Representation Only a small part of a binary image of 256x256 pixels is shown The shaded words are not part of the actual image but define a border around the image The word with index 19 is magnified to show the 32 pixels within one word Any image representation has to deal with the fact that structuring elements positioned at the edge only partly overlap with the image In the algorithms pixels not within the actual image are set to either 1 or 0 depending on the operation Bitmapped binary images 11 5 Image 2 1 Library User manual Note however that the border around the actual image which is also stored in memory makes border checking not necessary for structuring element pixels with coordinates k l satisfying Ikl lt 32 and Ill lt 1 In the algorithms described in this chapter we assume that Ikl lt 32 necessitating border ch
45. image do the following select Image Conversion isodata_threshold click OK select Arithmetic invert_im click OK This produces a binary image note the type b2D Let us first demonstrate some filters on binary images from mathematical morphology select Image Morphology erosion3x3 2 When you give commands one at a time there is no difference between Dismiss and Stop but if you were running a program there is Dismiss skips the erroneous command but still executes the rest of the program while Stop halts the execution of the program completely Getting started 3 17 SCIL_Image 1 4 User Manual choose Input Image B choose Output Image B click Do It The image in window B has changed slightly A more pronounced effect is achieved by repeating the command a few times This is the reason we used Do It rather than OK it leaves the dialog box visible click Do It 4 more times Restoring the rough outline of the shape is achieved by using the dilation operation 5 times select Image Morphology dilation3x3 choose Number of iterations 5 click OK This should give you a sense of the speed of the implementation By the way while we are in the morphology menu a transform in SCIL_Image related to morphology is the distance transform select Image Morphology distance choose Input Image B choose Output
46. in combination with image processing commands Image creation commands are a group of commands that help you create and destroy new images and regions of interest in images Getting started 3 39 SCIL_Image 1 4 User Manual Various image utilities contains some commands that occur in other menus as well but are often used in combination with image processing commands It also has the important copy_im command to copy images Display Mapping gt v auto_display set_clut clut_by_name bin_disp_colors create_display natural_window_size set_window_pos set_window_size set_display_slice next_plane set_dither_mode set_display_mode set_sigmoid_shape stp 2_planes dir_maximum stereo_view display_image part_image_display Figure 3 22 the Display menu This menu contains commands that manipulate the display of your images The commands can affect size location grey values and colors of the images Options dialog_options v auto_display bin_disp_colors Figure 3 23 the Options menu This menu permits you to tailor the interaction of SCIL_Image dialog boxes display reaction on mouse clicks in images etc to your taste 3 40 Getting started SCIL_Image 1 4 User Manual Arithmetic Int_based Float_based Complex_based Color_based Log_based Trigonometric Exotic ee a ee a a add_im sub_im fmul_im div_i
47. in the dialog history window TOGGLE Advanced SCIL_Image 5 7 SCIL_Image 1 4 User Manual The argument of the command toggles between 0 and and the command is executed directly without generating a dialog box The start value depends off the argument depends on the optional CHECK keyword following the TOGGLE Using this keyword on commands that have more than one argument results undefined behavior CHECK The command argument is on by default Used for the TOGGLE keyword Using this keyword on commands that have more than one argument results in undefined behavior DISABLE The command is disabled in the menu it can not be chosen DIRECT The command is executed immediately No dialog box is generated To indicate this no ellipses are printed after the command name DIRECT The command is executed immediately But because the command itself generated some kind of dialog box the ellipses are printed after the command name MANUAL name OPTIONAL The text which describes a function s syntax and behavior is normally stored in a file named after the function with the suffix man If this is the case the MANUAL field does not need to be specified However for a related group of functions it can be useful to group the explanations of those functions in one text file Rather than copying the file several times the MANUAL option can be used to identify the file which contains the explanation for the group The file m
48. int int int int int ney Tat int New image types 12 15 Image 2 1 Library User manual BYTE 1l_byte_read_masked_part BYTE BOOL_MASK int int int int int mt At int ant void l_byte_write_masked_part BYT int Git int aint Lit int int endif endif BYTE_2DP_H E W K E BOOL_MASK int int int The following is a source code listing of the low level functions used in the sample code in this section include lt stdlib h gt include lt stdio h gt include lt string h gt include image h include im_error h include im_infra h include roi h include byte_2d h include byte_2dp h BYTE l_byte_read_part_image BYTE ip int sx int sy int sz int width int height int depth int sw int sh int sd BYTE sp dp pte int Vi Zi im_begin_func 1l_byte_read_part_image if ptr BYTE malloc size_t long width height depth sizeof BYTE im_report_error l_byte_read_part_image IE_NOMEM couldn t allocate memory for buffer return NULL for z 0 z lt depth z for y 0 y lt height ytt sp ip zt sz sw sh syty Sw Sx dp ptr z width height y width memcpy dp sp width sizeof BYTE im_end_func 1_byte_read_part_image return ptr void l_byte_
49. lookup tables that can attached to the images name contents EMPTY_LUT all entries black BLUE_LUT a blue scale GREEN_LUT a green scale CYAN_LUT green and blue scale mixed RED_LUT a red scale MAGENTA_LUT red and blue scale mixed YELLOW_LUT red and green scale mixed GREY_LUT normal grey scale LABEL_LUT scale that simulates label display MULTLLUT primary colors each in a specified bit OVERLAY_LUT_1 8 specified bitplane red FALSE _COLOR_LUT false color table Table 9 2 pre defined color lookup tables Custom tables can be made by using the create_clut function filling the r g and b table with desired values After that the display interface must be told that the contents of the clut changed This is done by publishing aSPB_CHANGED message to the super clut object Programming with Image 9 11 Image 2 1 Library User manual void sclut CLUT clut sclut get_super_clut creating an filling the clut spb_publish super_clut SPB CHANGED clut Dynamic adjustment Pre_op Post_op Frequently in the course of image processing routines are called which produce an output image of a different type or size than that of the destination image In Image some tools are provided to perform the appropriate type and size conversions The same tools handle ROI processing By using them the programmer is freed from converting images directly or from having to be concerned with whether an inp
50. n aka ing n n Ben n r ach new image by handle_super_im and prints information about publishes from this image iA void handle_images void image EA3D SpbARI im_area image_output IMO_OUTPUT void id Image Handler int mess void data void cl message d data d n magepointer to s n mess data if is_image image image_output IMO_OUTPUT I ImageName IMAG switch mess case SPB_CHANGED if im_area data E image image_output IMO_OUTPUT 13 6 Creating an Imag e Application Image 2 1 Library User manual Rectangle changed d d d d d d n im_area gt x im_area gt y im_area gt z im_area gt width im_area gt height im_area gt depth break default break This function detects if new images have been made and subscribes the handle_images function to each new image void handle_super_im void sup_im void id int mess void data void cl image_output IMO_OUTPUT Super Image message d data mess data if is_image data image_output IMO_OUTPUT Data is Imagepointer to s n ImageName IMAGE data switch mess case SPB_ITEM_ADD subscribe to each new image spb_subscribe data NULL handle_images OL break case SPB ITEM DELETE unsubscribe from images that are destroyed
51. name of the command we are constructing add_one The overloading involves a function overload_func which can perform that selection on the basis of data we have given it Change the file once more to correspond to the box below the function g_add_one is unmodified and not shown fully and save it add_one c 3rd overloading the operation include image h include im infra h int func IMAG Gl IMAGE add_one IMAGE in IMAGE out func overload_func add_one in if func return NOT_OK return func in out f_add_one IMAGE in IMAGE out float indata outdata long num if pre_op in out ADJUST F_2D_SPEC FLOAT_2D return NOT_OK indata ImageInData in outdata ImageOutData out num ImageSize in while num outdata indatat t 1 0 return post_op out g_add_one IMAGE in IMAGE out SCIL_Image still needs to be told that the command name add_one has been overloaded to floats and integers To do so make a new file add_one ovl in the directory scilimage14 prog overload The contents should be add_one ovl overload file for add_one ABLE g2d GREY_2D G_2D_ SPEC 2 add_one g_add_one ABLE f2d FLOAT_2D F_2D_SPEC 2 add_one f_add_one 3 32 Getting started SCIL_Image 1 4 User Manual Such a file
52. not be removed during a SCIL_Image session Now the keyword permanent is available for this type of variables The ANSI C keyword const is recognized but not implemented the keyword is ignored The C Interpreter 4 17 Chapter 5 Advanced SCIL_Image This chapter describes the advanced use of SCIL_Image It shows how to add new commands to the system and how to change the user interface This section is rather technical more general topics concerning SCIL_Image such as the Direct Command Mode and the Program Mode can be found in Chapter 4 Read this chapter if e You want to add new commands to the system e You want to add help information to the on line manual e You want to change the user interface Do not read this chapter if e You have not already read Chapters 2 and 3 and preferably also 4 SCIL_Image 1 4 User Manual Adding New Functions to SCIL_Image This SCIL_Image version contains many basic image processing programs For your own applications you probably need special commands which may or may not be composed using those basic programs Very likely you will want to build up your own library of programs that are useful for your applications SCIL_Image allows you to do this in two ways One is to extend SCIL_Image with interpreted programs as we discussed in Chapter 4 The second is to create a completely new version which includes your own compiled functions You will be able to reach those functions f
53. occurs when the output of an operation is directed to the input image At several points in the remainder of this chapter examples of commands are given The user is encouraged to read this chapter while running SCIL_Image and try the examples All examples are given as command lines Most of them can also be issued using the menu as described in Chapter 3 Please note that in some examples not all parameters are specified in the text but are set to their default values by the command expander The reader should refrain from specifying the missing parameters because the resulting image type and contents of an operation may be assumed in subsequent examples 6 2 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual Image infrastructure Invalid operations In the course of processing an image various errors may occur For instance the specified function arguments may be of the wrong type or out of range A specified input image may be of an inappropriate type for the operation or may not even exist When errors of this type occur SCIL_Image pops up an alert box to warn the user Before continuing interactive work the user is required to respond by clicking the Continue or Stop button in the alert box To see how an alert box looks type the following command lines readfile trui threshold A B threshold B C The second threshold command request the system to perform the threshold operation on t
54. of the given class are accepted In the max field extra choices can be given presents a list with available histogram objects see Histogram objects on page 6 26 The min field is discarded the max field can be used to specify extra choices presents a list with available color lookup tables see Display lookup table on page 6 7 Advanced SCIL_Image 5 15 SCIL_Image 1 4 User Manual Creating a New Compiled SCIL_Image Version The description of the CDF files permits you to add your own non compiled programs to SCIL_Image even making them accessible from the menu However they will still be interpreted and are therefore slow and not fully accessible to other programs or functions or to other users You may want to make a new version of SCIL_Image that contains functions that are important to your applications in a compiled form Prepare write and possibly debug the program s in C that you wish to install in SCIL_Image Let us use as an example a series of functions to generate various types of random noise This means that you have to prepare the following documents My_sre directory noise cdf your command description file CDF noise ovl your overload file noise h this file contains your constants noise c this file contains your functions The rules for composing the noise cdf file are given in Making a Command Description File CDF page 5 3 The rules for composing the noise ovl file are giv
55. of the in1 image In the tables that contain the overloadable functions for each image type add_im is an entry and is accompanied by a function pointer to a type specific function In the table for GREY_2D images this pointer points to the function g__add and as a result of that this function will be called g_add inl in2 out Parameter checking and image adjustment The function g_add is the second layer include image h include im_error h int g_add IMAGE inl IMAGE in2 IMAGE out im_begin_func g_add 9 28 Programming with Image Image 2 1 Library User manual if pre_op inl in2 COMPARE G_2D_SPEC G_3D_SPEC OUT_AS_IN pre_op inl out ADJUST G_2D_SPEC G_3D_SPEC OUT_AS_IN return im_report_error g_add IE_PRE_OP 1_g_add inl in2 out if status post_op out lt IE_OK return im_report_error g_add status im_end_func g_add return IE_OK This second layer is used for image type and size adjustment as well as general parameter checking The functions pre_op and post_op which test the input images and adjust the output image are discussed in detail in Dynamic adjustment Pre_op Post_op Here the first call to pre_op compares the images in2 and in1 and produces a warning if either the dimensions or the type differ The second call to pre_op will adjust the image out to be
56. operation the type and or sizes of the image may have to be changed explicitly instead of letting the infrastructure handle it We pause here to remind the reader of the difference between an image and a display window The data is stored in an image which has a size and pixel value range depending on the type and declaration The display window is a completely separate entity acting as a viewport on the data with an independent size Creating and destroying images When starting SCIL_Image four images are created If more images are needed they can be easily created To create a new image with the name fifth_image issue the command make_image fifth_image 6 8 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual An image name may be any set of ASCII characters but may not contain white spaces A new image has popped up on the screen carrying the name fifth_image of type g2d and with the dimensions 256 by 256 This image can now be used for processing For instance invert_im A fifth_image Destroying an image is accomplished with destroy_image fifth_image Changing image sizes Changing the size of the image will destroy its contents To change the size of image A change_image_size A 128 128 Image A now has dimensions 128 by 128 Each pixel in A has been set to zero Changing image types To change the type of image C set_im_type C GREY_2D Image C is now a grey valued 2d image
57. pair for image type GREY_2D is add_im g_add e Every time that the keyword TABLE is encountered in an overload file a new table will be created This way multiple tables can be specified in one file e When the overload c is created it is not checked whether a name of an operation is already in use If that happens the last occurrence of the name and the corresponding function overrules the previous one s Note that in the overload file for GREY_3D images the message function pair for the add_im operation is the same as the one in the overload file for GREY_2D images This is because the g_add function can be used for both GREY_2D and GREY_3D images Overruling the default implementation It can occur that a user of Image is not satisfied with the implementation of an image processing operation or simply wants to experiment with a different implementation For instance if the default implementation of a filter for GREY_2D images is not suited for the specific images used the Image routine can be overruled by a different one This can only be done if the new function is called with the same set of parameters as the old one because the generic level function is unchanged If the parameters differ a complete new function must be implemented using a different name To overrule the Image implementation of a function the message function pair rule for the new function must be located later than the rule for the d
58. see Getting started chapter 3 More information on the image processing functions is given in Introduction to Image 2 1 chapter 7 and further The best way to get acquainted with the system is however to use and try functions on your images To enhance the flexibility of the system functions have parameters and options to be filled in by the user A parameter may be the symbolic image name a code to change the behavior of the operation a value used as a selection criterion etcetera In selecting the image processing functions from the menu the user is assisted by the system in the selecting parameter values 1 2 Outlines of SCIL_Image SCIL_Image 1 4 User Manual Usually a default value is available which should normally do the job Also with almost all functions help is available giving a written explanation to the purpose and method of the function and its parameters Using shorthand typing for application development When a menu option has been selected it is translated by SCIL into a command The command appears in the text window Subsequently it will be executed In SCIL_Image any menu selection is automatically translated to a written command inserted in the text window and executed only then This is done to serve the user as he or she becomes more experienced with the system and the image processing libraries More experienced users may find interaction via the menu system tedious They rather type comman
59. stand alone program without image display it is advised to execute this session in the SCIL_Image package as this shows the result of each statement directly on the screen First an image must be obtained In our example a file is read from disk thresholded and inverted so that object pixels are set and background pixels are cleared readfile cermet A 0 0 Obtain an image with objects threshold A B 128 Segment the image Not present in the stand alone Image Library see Interactive measurement on page 10 6 10 4 Analysis of Images and Objects AIO Image 2 1 Library User manual invert_im B B Object pixels set background zero Given a segmented image with objects and background the image must be labeled returning a list with objects In order to use AIO data structures must be defined using the standard C file inclusion mechanism include im_aio h AIO data structure definitions LIST o_list object Two list variables o_list list_label B C 8 0 Label the image return a list Since the labeling process creates a list with initial information including the area of the objects objects too small or too big to be of interest can be eliminated immediately In our example objects touching the edge of image C are removed as well Removal from the list is sufficient to never reference the objects again but for clarity we choose to erase the objects from the image as well FORALL object o_l
60. systems Therefore there is no direct way to read from the input or write to the output or error stream The image library provides an abstract error stack mechanism to facilitate the error reporting Also a special function for text output is used throughout the image library The textual output can be intercepted in any user interface by setting up a special output handler Publish and subscribe mechanism To safeguard a complete separation between the application code and the graphical user interface a publish and subscribe paradigm is used Every object publishes important events to itself Other objects interested in changes of that particular object may subscribe to it As a publishing object does not explicitly know its subscribers the separation between core application and user interface is guaranteed The advantage of publish and subscribe method is the possibility to extend the existing system with new user interface behavior without having to change existing code The image library has been implemented as a stand alone imaging library with publishing image objects This provides the developer with powerful Introduction to Image 2 1 7 3 Image 2 1 Library User manual tools to construct a user interface for the entire image library or allows to realize an end user application without any modification to the image library The structure of this manual In summary the remainder of this manual contains the following chap
61. t byte_mask 0x80 im_end_func l_byte_read_masked_part return ptr void l_byte_write_masked_part BYTE ip BYTE data BOOL_MASK mask int dx int dy int dz int width int height int depth int dw int dh int dd int clear BYTE dpy spi BYTE mp byte_mask int X Y Z line_offset im_begin_func 1l_byte_write_masked_part line_offset sp for data z lt depth z y lt height ytt z 0 for y 0 width 1 8 1 New image types 12 17 Image 2 1 Library User manual dp ip ztdz dw dh dyty dw dx mp BoolMaskData mask z height line_offset line_offset byte_mask 0x80 for x 0 x lt width x if mp amp byte_mask dpt sptt else if clear dpt t 0 else dpt Sprt if byte_mask gt gt 1 mptt byte_mask 0x80 im_end_func 1_byte_write_masked_part return Testing the image type If the above steps have been followed correctly a new version of Image has been compiled and linked which includes the byte_2d image type The following sample code can be used to test the new image type Please note that this is only a code fragment not a complete program it expects that the Image library is has been initialized include image h include im_error h include im_infra h include roi h
62. the so called BARRELSHIFT function The function BARRELSHIFT A i k 1 border is given in pseudo code in algorithm 2 We use the C notation lt lt gt gt to denote shifting of the bits in an integer ALGORITHM 2 Barrelshift Function Pseudo code for the Barrelshift function where A is the bitmap i is the bitmap index of the central pixels k 1 is the pixel position relative to these central pixels and border is the word returned when pixels outside the image are accessed BARRELSHIFT A i k 1 border DO j i 1 LineOffset IF 0 lt j lt MaxIndex The pixels are within the bitmap IF k 0 No Barrelshifting needed RETURN A j ELSE IF k lt 0O we are looking for pixels to the left of the central pixel RETURN A j gt gt abs k A j 1 lt lt 32 k ELSE we are looking for pixels to the right of the central pixel RETURN A j lt lt k 1 A j 1 gt gt 32 k ELSE the pixels are outside the bitmap Bitmapped binary images 11 7 Image 2 1 Library User manual RETURN border ENDDO The pseudo code for the erosion using an arbitrary structuring element is given in algorithm 3 In the algorithms the border effects when the structuring element is not entirely within the image are not accounted for Replacing the logical AND amp in algorithm 3 with a logical OR results in an algorithm for the dilation ALGORITHM 3 Erosion The pseudo code for the ero
63. the parent the functions copy_part_image and copy_masked_part are used The first one for rectangular ROIs and the latter for arbitrarily shaped ROIs The infrastructure uses the create and destroy functions to allocate the necessary data space and then the copy functions to put the data in The copy_part_image function is supplied with the position of the rectangle the sizes and the destination position which are all already checked for validity on a higher level As the function is also called from other functions it must include calls to pre_op to check and adjust the images it receives The input image must be COMPAREd with itself and the output image must be ADJUSTed to match the input image As the output image was ADJUSTed with pre_op it must also be given to post_op The function must return OK 1 if it was successful and NOT_OK 0 otherwise Its function heading and the calls to pre_op and post_op are int type_copy part_image IMAGE in IMAGE out int sx int sy int sz int width int height int depth int dx int dy int dz if pre_op in in COMPARE type_spec ImageTypelIdent in pre_op out out ADJUST ImageTypeSpec out ImageTypeldent in return NOT_OK the actual copying return post_op out For the byte 2d image type the implementation of the routine uses two lower level routines l_byte_read_part_image and1l_byte_write_part_image see Lo
64. the time needed for the measurement is a point of concern it is better to apply a hierarchical classifier which depending on the state of a decision tree sequentially decides what feature of an object to measure next We therefore decided to support the measurement of individual objects rather than the measurement of all the objects in the image 10 2 Analysis of Images and Objects AIO Image 2 1 Library User manual In addition to the single object measurement the manner in which objects are stored is important for microscopical image analysis applications For instance in automatic screening for medical applications requests to store images of measured or classified objects for later inspection by a medical expert are common When building a classifier stored object images may be used to construct test and learn sets How the images are stored plays a role in the access times for these operations Another important consideration in microscopical image analysis is the style of user interaction In many applications the capacity to interactively manipulate the objects adds to the usability and functionality of the application It may be useful for example to point at an object of special interest to initiate a measurement or to correct for mis classifications We therefore support both automatic and interactive microscopical image analysis Based on the preceding discussion the AIO framework is designed to support microscopica
65. to manipulate Var_object structure The structure that describes a var_object is found in the include file image h typedef struct var_object_t int type type of data in this object void data pointer to the data int nr_channel number of channels per element int dimensions number of dimensions int dims V_O_MAX_DIM dimensions int size total number of elements int bpelem bytes per element char object_name IM_NAMELEN name of the object char object_class IM_NAMELEN class of the object char comment pointer to possible comment VAR_OBJECT As with the IMAGE structure all members of this structure can be referenced with macros define V_O_Type op op gt type define V_O Data op op gt data define V_O_Chans op op gt nr_channel define V_O_Dims op op gt dimensions Programming with Image 9 37 Image 2 1 Library User manual define V_O_Dim op dimension op gt dims dimension 1 define V_O_Size op op gt size define V_O_ElemSize op op gt bpelem define V_O_Name op op gt object_name define V_O_Class op op gt object_class define V_O_Comment op op gt comment for the total number of bytes in the object define V_O_Length op V_O_Chans op V_O_Size op V_O_ElemSize op e type is the type of the data the var_o
66. trui im_a 0 0 threshold im_a im_b 128 bm get_bool_mask im_b convert im_a im_c BYTE_2D im_d create_image D BYTE_2D 256 256 1 roil roi_define roil im_c 0 0 0 256 256 1 bm convert roil im_d FLOAT_2D writefile im_d result3 ICS_F We have described a very limited implementation of the byte image type To do something useful with the new image type additional image processing operations should be added in a similar way Defining an image subtype Image subtypes can be implemented for a number of reasons The difference in the interpretation of the image data is one reason but sub typing can also be used as an organizational tool It is possible to design a set of operations that is targeted to a specific processing domain such as microscopical images By limiting the operations that can be performed on specific images their specialized nature can be emphasized and the user can be aided in processing them The implementation of an image subtype is similar to that of a new image main type described in the previous section Some definitions need to be made and the service functions must be made available The difference lies in the fact that an image subtype can make use of functions already implemented for another image type In this section we describe how to implement an image subtype The standard Image Library image type LABEL_2D is a subtype of the GREY_2D image type a
67. using the command readfile Most images will have to be read from disk before they can be processed Because there are several ways to create an image cameras microscopes etc the data format must be known in order to read it The dimensions and image type must also be known For this reason several standard image file formats have been developed Currently SCIL_Image supports the following file formats for images ICS format TIFF format JPEG format TCL format AIM format Because all SCIL_Image image types 2 D and 3 D can be stored using theICS format we use this as our primary format We recommend this format to store images produced using SCIL_Image The ICS format With the ICS format image data is stored in two files One file contains the actual data of the image and has the extension ids The other file contains header information such as image type and dimensions and has the extension ics This header file is an ASCH file which can be read and edited using standard tools Aside from information about the type and dimensions of an image data regarding the resolution of the image date of creation and other pertinent details may be stored in the image header file See the appendix ICS file format for a full description of this image file format The TIFF format TIFF stands for Tag based Image File Format and is designed for the interchange of digital image date The format claims to be independent of a
68. vas cuanse duces Hes seasze seauastes oobacte esas accdcodsteds cone taseoes 6 24 Behavior OF var objectsacrenoe e i ei E tains Maes ae eewdacang 6 24 Datatypes of var Objects secina esena eniin ei ea ikos 6 25 Examples usins yar Objects roians era e a nat eat TAR A E 6 25 EAPO Le Ue a cel Seat ec cuidse sas code TA E sagntdend esoeandenteaceasacowees 6 25 Example 2 pee ae en ee AER eee NE eR COR Ee eer ee 6 26 Example snan etree Speer E re Corse CR fe cnt rerat fre terttnr 1 E aA 6 26 Storage of var Objects on disk onenean a gah teat 6 26 Histogram Objets arein oe a aa A AN A e a e 6 26 Chapter 7 Introduction to Image 2 1 essssooccessssoccecesosocceccssooeesessssesese 7 WV Hae 1S lene 2s EEE E E E E easton 7 2 PIAS CN AS TU CUS ss ayy i eaa Oa EEE eet Be te Maced ake sae i E 7 2 imase type Sss ta ccaeehacihacus ces leatesans aa aa oe waacane toaeet wa eae e eee 7 2 Advanced and extensive set of image operations eeeeeeeeeeeeeseeeceeeeeesteeeeeaes 7 3 F st implementations ss ate gee nataaucalas outa veaseauuisaedgns sesprased eoeonddatanceneasonees 7 3 Abstract error and I O handling 5 5 5 0 52 ui aut ntica cacy ves eas aden ands tue Sela ceson ty cetevar ences 7 3 Publish and subscribe mechanism 2 25 2 ci02 ec tsiectasstieeseniciea ceased 7 3 Thesstructure of his mantal 25 alaiee hes te a Reale e aie inde gehts eet 7 4 Whe need for Managed ela seats r a e E eos a teecaetaeeees 7 4 Str ct reof the Image NOTARY aoee so anal a
69. well as the width of one cell Functions are available to create destroy and resize the histogram objects Histogram structure The definition of the HISTOGRAM structure located in the include file image h is typedef struct histogram_t unsigned long hdata unsigned int nr_ chans unsigned int nr_dims unsigned int dims V_O_MAX DIM char name IM _NAMELEN char comment double lbin_median V_O_MAX DIM double hbin_median V_O_MAX_ DIM double bin_width V_O_MAX DIM void futl reserved void fut2 for void fut3 future void fut4 use HISTOGRAM 9 40 Programming with Image Image 2 1 Library User manual For all used fields in the structure a macro has been made for accessing the field gt hdata gt nr_chans HistogramData hp hp HistogramChans hp hp HistogramDims hp hp gt nr_dims HistogramDim hp dimension hp gt dims dimension 1 HistogramLBinMedian hp dimension hp gt lbin_median dimension 1 hp hp hp hp HistogramHBinMedian hp dimension gt hbin_median dimension 1 HistogramBinWidth hp dimension gt bin_width dimension 1 HistogramName hp gt name HistogramComment hp gt comment e hdata is the pointer to the actual histogram data The type of the data is unsigned long e nr_chans is the number of channels of the histogram data When the image dat
70. within a valid range 0 number of bins Programming with histogram objects Several functions are available to create destroy and view the histogram objects They are create_histogram creates an empty histogram object destroy_histogram destroys a histogram object histo_data gets the histogram data from an image dump_histogram show the histogram data in ASCII show_histogram_info show textual information of a histogram When using the histogram objects you may need some additional functions for inspecting and manipulating the histogram objects 9 42 Programming with Image Image 2 1 Library User manual histogram_comment attaches a comment string to a histogram object histogram_by_name gets the pointer to a histogram objects using its name is_histogram checks if a pointer is a valid existing histogram object and returns true or false histogram_ok checks if a pointer is a valid existing histogram object and pops an alert box if this is not list_histograms lists all existing histogram objects copy_histogram copies the data of one histogram to another When it is necessary to copy the data in the histogram object to an image or var_object or vice versa these functions can be used histogram_to_image copies the histogram data to an image image_to_histogram copies image data to an histogram object histogram_to_var_object copies the histogram data to a var_object More detailed information on the function
71. x y maximum of x and y float min x y minimum of x and y float pow x y x to the power of y float rnd random number integer sin x sine of x float sqrt x square root of x float tan x tangent of x float Table 6 4 functions in eval In addition to the operators and the functions in eval two keywords are available for special use The words xx yy and zg present a pixel s position in an image so to fill an image with a grey ramp eval C yy The pixels in image C now all have a value equal to their y position in the image The last operator in the table is the query operator This operator is used for testing and taking different actions depending on the result of the test To use the operator three different parts need to be specified The first part before the question mark must be the test the second part after the question mark and before the semi colon is the part that is executed if the test is passed true The third part after the semi colon is executed if the test failed This construction should be well known to C programmers readfile trui B eval C xx lt 128 A B Any of the three parts may be any expression accepted byeval so statements of the following sort are acceptable eval D xx lt 128 A 128 0 255 B The Image 2 1 library in SCIL_Image 6 21 SCIL_Image 1 4 User Manual Storing images on disk The images used thus far were read from disk into memory
72. 0 Discussion and C Onc lUStON Sc ce2sce0cc28234 soassva coencttaicceanienscetd di soacsangaetnigeccaeanie teed 11 10 Literaturen tte etor acca Scat a aga sakinety edged dotted eal tattuaat styled atdae a vom ttes 11 11 Chapter 12 New image types ssssssssooeccssssococcesssoccecessosoccecsssooeecesssosoeeeee 12 1 Implementing a new image type sicisiscgcsccassaccesasesesacesescaceesdecadenspavsdvabeusaccesnnccd tscavecees 12 2 D fines and strct reS iersinii tag ani iii 12 3 Creation and destruction eessseseseseessesesesressesereseessesetsstessesstertessetstsseeseeseeseresee 12 4 Copying a part of the image essseeeseeesseresseseresresseserssresstsserrressersresreeseeseeserssee 12 6 Displaying the image nsss cepted lratene ae eee ceale n isa Waa 12 8 Image type informatio eseese sair areena oad tas cpestameron nets 12 11 Conversion to Other image ly pessoa GOES 12 12 Th er TTA xi asa ase de ceeds esa te EE eee ee see eae nda 12 14 DIF LOY CLS a cect aae stat eced dh seat E Wie stat actase S iaata Gaeta 12 15 Testing thesia ge ty Pe terisi era tates hain oa eG iet inde eke eet 12 18 Defining an image subtype ica eenn iiir tee Carat dea 12 19 Chapter 13 Creating an Image Application cccssssscsccsssssssccsssesees 13 1 ETAn 1AY 4 A 6 1 EEE et Posen eer pen ArT cor et fro Re POT aOR Re atSeeT oar Reet yeh er er em 13 2 Error ani ges 0 Basak a ie Paces ha gale Sete Rees 13 2 Text Output handling ac ects acct ea ir a e
73. 1 This function results in a list of information about the object appearing in the command window For instance the object s name class number of dimensions sizes and pointer values will be displayed 6 24 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual Datatypes of var_objects The basic standard C types may be stored in var_objects Further the SCIL_Image type PIXEL used in grey valued images may be stored in a var_object A complete list of all types is type name description PIXEL data type of GREY images 16 bits char character one byte short short integer 16 bits int integer 32 bits long long integer 32 bits float floating point number 32 bits double double precision floating point number 64 bits Table 6 5 var_object data types integers in C may have a different number of bits depending on the system and compiler used The value used here is the number of bits used on most platforms Examples using var_objects The var_objects are used in several different operations in SCIL_Image Below we show some examples Example 1 A method to display a 3D image in 2D is to display the maximum pixel value of the pixels in corresponding locations in the various slices In SCIL_Image a function is available that can determine either the maximum value of the whole image or the maximum value along an axis The result of this can be printed on the command window or stored in
74. 10 TAGE TOMA Cas are dees ins candidates acanignatete Sae T EESE tana 4 10 logon lt filenames logoff iis cotta that teeing einai e a a a 4 1 macro i v lt macrofile gt oo ec ccsecccccccccessessnsecececececeesessesecececeeeesersseeeseeeens 4 11 TAVAT Sao es a i e EE E E dee E E le a E eee 4 12 run ates hannen once I N Met nemeaeea teers 4 12 TINTS COMA gt an oe od E A tended e EE ERE E eA an E an 4 12 theinterr pt PAUSCH ES PEAK 55s eorn e ize ee aa ee a aet 4 13 help facilities 2 4 ca e eal Bie ese eae ete a tee E E Rae ees 4 13 Ctrl H lt selection gt Ctrl Enter lt command gt lt pattern gt eee 4 13 Errors Warnings and Diagnostics 22 2cauistietlem ales eee aati eee aia ad 4 15 Features of SCIL_Image s C interpreter say scagcay ase cadyaaseansaaedenshassexeodenoeendsntaeceosaaosecs 4 16 Chapter 5 Advanced SCIL_IMage eoessssecccesssscccecesosocceccssoosecsessssesese 5 1 Adding New Functions to SCIL_Image eeeseseeeseeeesseseesressrsrrssressrsererrersessrssressesee 5 2 Making a Command Description File CDF sssssessesssesessessessresressessresrersessrerreesesee 5 3 comment SINE YE oer gs na a e A ac A N Te 5 4 MEN CNY en a erat aE R E E e EE E E A Jan E a 5 4 GATS ALS S D A EEEE EE EE EE E N EEE E E 5 5 variabler entry certe Pies ao EEE E E E KE a EEE ee E ids 5 5 command ONLY ics sc ccrcchtacaccsanteartaaent a ea a ae a ai e RS 5 6 SCIL Image Special Pype Skeci ra a a a een a dal EK 5 10 Very Advanced SC
75. 2 1 Library User manual pre selection can be made based on fast measurements The time consuming measurements need only be applied to those objects which survived lowering the classification costs Object manipulation To manipulate the objects basic utility functions are provided They are removing an object from an image copying an object from one image into a randomly positioned window of another As objects are referred to through their associated data structure measured results can easily be accessed A feature can be measured on every object by applying the measurement to each of the objects in the list A selection criteria may be applied while traversing the list Image Silo An image silo package enables objects or group of objects to be stored in image silos for later retrieval and possibly further analysis Furthermore for presentation purposes routines are provided for laying out groups of objects in the form of composite images Direct manipulation Event driven interaction based on mouse and keyboard input is supported in AIO AIO enables simple interactive selection of objects by scanning the object list and returning the object at a given location Direct manipulation and user interfaces for image object manipulation can be written based on it An AIO sample session In a number of examples some of the concepts of AIO are demonstrated Although this the code fragments in this section can be used in a
76. 256 top left b Cermet 256x256 top right c Musicscore 1024x244 bottom Pixelwise Logical Operations The pixelwise logical operations on bitmapped images are much faster than the equivalents working on pixel mapped images This is due to the fact that instead of operating on one pixel value at the time we operate on 32 pixels in parallel There is also no need to isolate the pixel bit from the grey pixel value fetched from memory Table 11 1 shows the experimental results for a 256x256 binary image The algorithm for the pixel mapped images is denoted as PIXELMAP The table also shows the results for the SIP machine BITMAP PIXELMAP SIP Operation ms __ ms _ ms AND 2 0 87 3 93 0 OR 1 7 87 3 93 0 EXOR 1 7 115 0 91 0 INVERT 1 3 74 0 N A COPY 1 3 74 3 46 0 Table 11 1 Pixelwise Logical Operations Timing results for the AND OR EXOR INVERT and COPY operations Bitmapped binary images 11 9 Image 2 1 Library User manual Morphological Transforms The bitmapped implementations of the erosion are compared with two existing software implementations and with the special purpose image processing machine SIP The most simple method to implement erosions and dilations using structuring elements of size 3x3 is to use look up tables A table with 512 entries holds the results for erosion dilation etc for all possible configurations in a 3x3 neighborhood Transforming the image thus boils down to one ta
77. 2600 AD Delft The Netherlands The Netherlands Stieltlesweg 1 2628 CK Delft The Netherlands SCIL_Image version 1 4 Page 1 of 268 8 26 11 69 in 4 Figure 3 26 the acrobat reader Navigating through the manual can be done in several ways using the cursor keys the scrollbars the buttons in the toolbar or clicking the bookmarks in the left pane Extended help on using the reader is given in the Help menu of the reader itself Reference manual The reference manual opens as shown below at the index page Getting started 3 43 SCIL_Image 1 4 User Manual Acrobat Reader scil_ref pdf olx WIR Fie Edt View Tools window Help 18 x afen Ma a gt rt ae olam a D DO Reference Mant lt D Index SCIL_Image 1 4 Reference Manual D D 8 Da Dc DD DF DE Os DH D Da Dk DL OM DN Do DP Da QR Ds T Duw Dy Dx my Index add_applic_e add_applic Figure 3 27 the reference manual index To find the reference page of a function search for the function in the Index and then click on the page number behind it within the red square The reader then jumps to the page on which that function is described As shown below with the readfile function Acrobat Reader scil_ref pdf x HA Fie Edit View Tools Window Help 18 x SfE m 7 ale lal il gt gt BCT Al DD Reference Man SCl mage l4 Reference Ma
78. 2d_conv_from_common IMAGE im int n COMMON_LINE com_line GREY_2D_ IMAGE g_im double d_ptr long l_ptr Programming with Image 9 35 Image 2 1 Library User manual PIXEL pix int type register long npix register int chan_offs im begin func g_2d_conv_from_common g_im GREY_2D_IMAGE ImageOut im type com line gt type avptr com_line gt data l_ptr com line gt data chan_offs com_line gt nr_channel npix Grey2dImageWidth g_im pix Grey2dImageData g_im pix com_line gt y npix r F if com_line gt z 0 if type COM_LONG while npix gt 0 pixt l_ptr l_ptr chan_offs else if type COM_DOUBLE while npix gt 0 pixt d_ptr d ptr chan_offs else if type COM_LONG while npix gt 0 if pix lt l2ptr pix l_ptr pixt t lptr chan_offs else if type COM_DOUBLE while npix gt 0 if pix lt d_ptr pix d_ptr pix d ptr chan_offs if com_line gt t gt 0 im_report_error g_2d_conv_from_common IE_NOT_OK return 0 0 stop can t handle this im_end_func g_2d_conv_from_common return 1 1 can handle more data 9 36 Programming with Image Image 2 1 Library User manual User specified conversion The routines that perform the ac
79. 3D_ 8 C_3D_SPEC 128 color 2 d COLOR_2D 9 COLOR_2D_SPEC 256 color_3d COLOR_3D 10 COLOR_3D_SPEC 512 labeled 2 d LABEL_2D 33 L_2D_SPEC G_2D_SPEC labeled 3 d LABEL_3D 34 L_3D_SPEC G_3D_SPEC Table 9 1 present image types Each type_ident has a unique value which is used to identify the image type The label type_spec can be the same for more than one image type as is the case for labeled and grey valued images They have the same type_spec because both the data representation and layout are the same for both image types Functions which operate on images of type GREY_2D are also capable of processing also those of type LABEL_2D GREY_2D is a class and LABEL_2D is a subclass of it Due to the interpretation of the data however not all operations that are meaningful on a GREY_2D image are valid on a LABEL_2D image and vice versa For example a median filter is a useful operation to perform on a grey valued image for purposes of noise reduction but performing it on a labeled image is of little value A table of available operations for each image type is maintained to keep track of which operations are admissible for that type The overloading mechanism is used to apply the appropriate function based on the image type Therefore functions which perform type Programming with Image 9 3 Image 2 1 Library User manual specific operations can be kept relatively simple since they need only consider the specific char
80. 5 18 Advanced SCIL_Image SCIL_Image 1 4 User Manual Compilation will take some time the help compiler will minimize itself during compilation and restore itself to normal view after compilation It generates the new scilimag hlp which makes your Help page reachable from SCIL_Image like that of any other command amp SCIL_Image 1 4 Windows Manual OF x File Edit Bookmark Options Help coments rder on Bie e NAME add_one add 1 to an image SYNOPSIS include user _def h include image h int add_one IMAGE in IMAGE out DESCRIPTION Add 1 to the image in and store the result int the image out RETURN VALUES OK 1 if successful NOT_OE 0 ifnot successful AUTHOR written by M Y Self 19970204 As is customary in Windows on line help the underlined colored terms represent links to other help pages or pop up windows when the underlining is dashed Advanced SCIL_Image 5 19 Chapter 6 The Image 2 1 library in SCIL_Image This chapter introduces some basic features of the Image 2 1 library as part of SCIL_Image Do not read this chapter if e You have not read Getting started chapter 3 Specifically it is assumed you know how to handle the menu amp dialogue boxes and you know how to work with the command line editor Read this chapter if e You want to process images in Image e You want to know more about image types e You have basic knowledge of ima
81. 6 bits 2 bytes per pixel are used Therefore the value ranges from 32768 to 32767 a total of 65536 values readfile house A If the pixels are examined with the use of the left mouse button it can be seen that every pixel has an integer value within the range 0 255 This is because the camera this image was made with only used 8 bits per pixel This type of camera is very common and therefore most grey valued images encountered use only 8 bits per pixel Why use 16 bits per pixel There are devices that create images with more than 8 bits per pixel There are also many operations on images with 8 bits per pixel which result in images than use more than 8 bits per pixel The Image 2 1 library in SCIL_Image 6 11 SCIL_Image 1 4 User Manual Usage of grey valued images The grey valued image is the most common type used in image processing because its data representation matches the computer s data representation The computer can therefore access the data quickly and easily The amount of information that can be stored per pixel is sufficient for the majority of purposes Examples of grey valued operations It is frequently useful to enhance the contrast in a grey valued image We illustrate two operations available in SCIL_Image for doing so readfile maan A show_histogram A contrast_stretch A B 2 98 show_histogram B equalize A C show_histogram C The histograms for all three images were created in separate windo
82. AGE_TYPE_FIXED ip gt in_descript gt type ImageOutType ip IMAGE_TYPE_FIXED ip gt out_descript gt type ImageInData ip IMAGE_TYPE_FIXED ip gt in_descript gt data ImageOutData ip IMAGE_TYPE_FIXED ip gt out_descript gt data The meaning of remaining fields of the type descriptor depends on the image type An effort has been made to keep them as similar possible to simplify the programming task An example of a type specific descriptor also in the file grey_2d h typedef struct grey_2d t int type Type Identifier TYPE_FIXED short data Pointer to Image Data TYPE_FIXED int lenx Image Width int leny Image Height int plane Plane specifier GREY_2D_IMAGE The other type descriptors can be found in the include files for the different image types grey_3d h float_2d eto In the general part of the IMAGE structure the members in_descript and out_descript are pointers to image type specific descriptors The reason for using two type descriptors is discussed in Dynamic adjustment Pre_op Post_op on page 9 12 Image Flags The flags field of the IMAGE structure is used to store some special treatment flags for the image Each flag occupies one bit of the 32 bit long word Currently these flags are defined READ ONLY bit 0 integer value 1 image is read only the infrastructure does not allow
83. B RGB model while image B reads col2d HSI HSI model Also the left mouse button positioned in image A and B will display the difference between the images XXX YYY R redvalue G greenvalue B bluevalue Compare this with the same image in B XXX YYY H h hhhh S s ssss_ I iiiii When processing color images an often used method that produces good results is to process each of the channel or just 1 or 2 separately To do so in SCIL_Image the channels must be stored in separate images and then processed using a grey value operation either in integer of float For instance reducing the intensity of the image in B set_im_type C float_2d copy_channel B C 2 0 eval c c 5 copy_im B D copy_channel C D 0 2 Usage of color images At first glance it may seem that recording and processing color images is a nuisance as they require a more sophisticated camera and a threefold amount of data In many cases however using a color image simplifies the image processing task Rather than having only 256 different grey values at one s disposal for every pixel the amount for a color image is much larger making it easier to group pixels on the basis of their pixel values only Color data processing with the color data structure gives the user access to one multi valued digital color image rather than keeping track of image data scattered over several single valued images one for each channel of the color model 6 18 Th
84. C interpreter In SCIL_Image version 1 4 the C interpreter does not have the full functionality of the Kernighan and Ritchie 1978 C language A small number of topics are missing some of which may become available in future versions of SCIL_Image Furthermore a small number of ANSI C features have been added or are recognized but further ignored to improve flexibility 4 16 The C Interpreter SCIL_Image 1 4 User Manual The main differences are ANSI C calling convention arguments are no longer promoted to int or double before they are passed to a compiled function no initialization of arrays and structures no goto statement and labels no bitfields octal bit pattern setting is only possible for single characters not in strings no comma operator is supported except in a for loop As in for i 1 j 1 i lt 2 i the ANSI C preprocessor construction if defined is supported the ANSI C preprocessor keyword pragma is recognized and completely ignored In the interpreter the preprocessor symbol INTERPRETED is defined The differences in scope rules are Extern only works for interpreted variables compiled variables can only be referenced by using the VAR keyword in the comfiles and rebuilding SCIL_Image see Making a Command Description File CDF Static variables are NOT supported In older versions of SCIL_Image up to version 1 2 a static variable was a global variable in the interpreter that could
85. CIL_Image you may wishto define your own image types You can do so but it is advised only to do so when you have the proper experience See Chapter 12 for details Binary mathematical morphology A complete and fast implementation of mathematical morphology routines is available These include routines for erosion dilation closing and opening using arbitrary sized and shaped structuring elements These are optimized implementations for circular structuring elements Grey value morphology operations are also included in the Image libraries Numerical analysis of objects in images When images consist of a set of small objects such as microscopical images of cells and grains or local details the AlO library provides facilities to measure object features Measurements can be made in automatic or interactive mode The list of such feature values is stored in an editable file Such a file may be converted with the aid of an editor into a file suited for a standard statistical package Outlines of SCIL_Image 1 13 Chapter 2 Setting up SCIL_Image This chapter describes how to set up SCIL_Image It tells you how to install the SCIL_Image software and how to set up your working environment Read this chapter if e You are setting up SCIL_Image e You would like to change your working environment Do not read this chapter if e SCIL_Image is working to your satisfaction SCIL_Image 1 4 User Manual Setting Up Step
86. Complex images COMPLEX_2D amp COMPLEX _ 3D ceceeeeeeeeneeeneeee 6 15 Data representation of complex images ceeesceceseeeceeeeeceeeeeesteeeesteeeesaes 6 15 Usage Ol complex TAGES sass 2 cee aczsiuetee oY sed doe soaue sheds sooact ta esas dera conv taseees 6 15 Examples of operations on complex 1MageS e ce eeeeeseceseeeeeeeeneeeneeeeeenees 6 15 Labeled images LABEL_2D amp LABEL_3D u 00 eeccescccessceseeeceetseccetneneeneees 6 16 Data representation of labeled images cece eseeeseceeeeeeeeeeseeesaecneeeseeeenees 6 16 Usaseot labeled iMa SES each Gates nase a Gadus aaupacedgns E Ea niaes 6 16 Examples of operations on labeled images eeeneeeeeeeseeeeeseresrseresresrseresee 6 16 Color images COLOR_2D amp COLOR_3D oo eeeceeeeeceteeeceeeeeeeteeeenteeeesaes 6 17 Data representation Of color images 0 00 0 eeeeeecesececeeececeeeeecesececeeeecsteeeesaes 6 17 Usage OD COLOR IMAGEN is a e aso aodute a a ges nctta aces ANS 6 18 Expression evaluation on images eval eee ceeeeeseesseceseceseeesneecsaeceseesseeseneeesaeens 6 19 Storing images on diskais ierra ensi eE Ea E aE TEE SESS 6 22 The TCS fomato aie a n a E a 6 22 The TIFF fomato Peters reer reer R ferns Petre re AA e a EIRA Ea oe 6 22 The JPEG formatore aai dee Ze EA E E E Mace dy eel E 6 23 The TCG f rmat r ol cee a E E S E NTS REE 6 23 The AIM Torin asc crt eed ea tended E ome eA tae Ea 6 23 Non image data var_objects ccs seins
87. Getting started 3 3 SCIL_Image 1 4 User Manual SCIL Main Window lolx File Edit SCIL Image Display Options Arithmetic Itools Window Help ET A g2D 2567256 Mm EJE B 92D 2567256 lel Ea C 92D 2567256 Ke F F F er a t Worksheet SCIL_Image 1 4findows ifr History readfile trui A 300 300 c Copyright 1991 1998 by University of Amsterdam Faculty of Mathematics and Computer Science Amsterdam The Netherlands License number 9110aa Owner SCIL_Image development Welcome to SCIL_Image 1 4fvindows readfile trui A 300 300 Figure 3 2 The opening screen of SCIL_Image The windows you see are all standard windows which may be moved around resized and scrolled in the standard Windows way as well as minimized The bar at the bottom of the SCIL Main Window has a status indicator on the left which should now read Ready The system is ready to receive its first instruction In this session we concentrate on display of images so first open an image Point with the mouse at the Image menu click and hold A menu will appear Move the mouse pointer while holding down the button to the item named I_O Another menu appears move the pointer to readfile and release You can actually also do this by clicking on the menus that appear rather than holding the mouse From now on we will abbreviate mouse action sequences Thus the above sequence becomes 3 4 Getting started SCIL
88. IL_Image New Types s ssssesesssesesesesssesseessersseeeseeesssressresseessee 5 14 Creating a New Compiled SCIL_Image Version eesseseesesrsssesressessresrersersreeresseeee 5 16 Adding On line Manual Files to SCIL_Image eee eeeeeseceseeeeeeeeneecnaeeneeenees 5 17 Chapter 6 The Image 2 1 library in SCIL_Image cessscssseoeee 6 1 PAO CUCHION P toatl es O cies coal oa cate ga teganddoc ieee ec tond Coameteai ess 6 2 Image mira CC Uae sa eas oa funded Sed ae as keene a aaron ede beaten iea 6 3 valid operati ns sie 3 ola cela oh sedan lect eaves asec e Aa Tie 6 3 Image display and window management ec eeseeseceseeeeneecseecneeeeeeeeaeeeaeens 6 4 Mo s BU ELOMIS siae iis ce naced aa e a roads aa eSa Ta stacked 6 4 The title bar of image windows esseseessesseeseeesesressesereseessesrrssressereresresseseee 6 4 The left mouse button ir ceana a a eiaa ioes 6 4 Displayi e the imagem niio e E E bedi de Nee gh oe EE ee 6 5 Displaying SID AMASES oas tscaciti as a aE aA E E EE a AENT A 6 6 The treht mouse DUTON sisii eae a g a a 6 6 Changing a Window S SiZe cessoci nienia ieni a ie KiS i a ei 6 7 Changing a window s position sessseeeseseesseeesesresstserestesseseresresseserssressesee 6 7 Display lookup tables sensia ssaa ir n eao e er SES 6 7 Image management ist uenis a n a R teed R 6 8 Creating and destroying iMages sessesesseesssesseesseessertsseetsseessresseesseeessees 6 8 Changmg Mase
89. Image C click OK A distance image will appear in window C Every point in this image has a grey value proportional to the distance of that point to the closest object You could also do the erosions by thresholding the distance transform image The binary image in window B contains many objects SCIL_Image can perform measurements on these objects select Image Single_Objects measure choose Interaction YES click OK 3 18 Getting started SCIL_Image 1 4 User Manual Imeasure Pile Es Grey image A CB CC CD Binary image CA B CC CD Garbage level fo Shape 2N GG LCR BEND MIN XMAX YMIN Density MV GREYYAL J TRANSMIS f OD Print results YES C NO Store in file Browse Doit Dismiss Hep Figure 3 12 The Imeasure dialog box A new window called labelled_image appears with the objects in various colors gt labelled_image I2D 256 256 imi Ea A Figure 3 13 A labeled image click at individual objects in this window In the Worksheet window a list of feature values grows reporting on the shape measurements you selected To conclude measuring point at the labelled_image window hit the lt Enter gt key Getting started 3 19 SCIL_Image 1 4 User Manual This is the only way you should conclude the measuring You could have written the measurements to a file using the Store in file option of the measur
90. L_Image commands a macro chain loads file with a fully correct C syntax a C program rmvar The rmvar command clears the C interpreter s memory it removes the old programs including all its functions variables structure descriptions type definitions and pre processor defines Note that variables defines and typedefs are also removed whenever aload or chain command is given rmvar list run args The run command optionally with arguments is used to execute a previously loaded file The run command first declares all global variables and then calls the function main with the specified arguments if any time lt command gt The time command records the time it takes to execute the specified command This is quite useful for benchmarking your commands For example int t 1000 time while t time 0 250 4 12 The C Interpreter SCIL_Image 1 4 User Manual the interrupt Pause Break To interrupt a running program hit Pause Break This indicates to SCIL_Image that the execution of the current program should be stopped SCIL_Image will then return to the Direct Command Mode The status bar indicator changes from Running to Ready help facilities Ctrl H lt selection gt Ctrl Enter lt command gt lt pattern gt In SCIL_Image information on commands or functions can be obtained in various ways and to various extent Ctrl H pops up the on line Help facility f
91. MAGE image int n COMMON_LINE com_line e image is the input image e nis number times the function has been called when first called n 0 e com line is a pointer to the COMMON_LINE structure used in the convert operation Programming with Image 9 33 Image 2 1 Library User manual The first time the source function is called n 0 it must allocate memory to hold a line of data from the source image and fill the structure COMMON_LINE to describe the format of the data To fill the structure the function set_common_line can be used see on line manual The allocated memory is freed by the function convert when the conversion is completed The fields y and z are filled each time the source function is called with the position of the line in the source image In a 2 dimensional image y n In 3D images y starts from zero each time the z position is raised z also starts from zero and ends at the value depth 1 The fields hint_min and hint_max must be set to a value either the exact minimum and maximum value in the image or a good estimate thereof In the Image implementation these values are set to 0 0 and 255 0 for all non binary image For binary images the values are set to 0 0 and 1 0 Every time called the source function fills the memory with the next x line of the source image The return value of the function is used to indicate whether more lines follow the present line If there are
92. MAGE DATA SPACE p gt type BYTE_2D data gt lenx width leny height allocated space width height bytes The destroy image function is responsible for freeing the space allocated by the create image function In general it will free the data space occupied by the image and then the type description structure The function has one argument a pointer to a type descriptor and no return value Its mandatory function header is void type_destroy type descriptor pointer The byte_2d_destroy_image function void byte_2d_destroy_image BYTE_2D_IMAG Gl bp im_begin_func byte_2d_destroy_image if bp if Byte2dImageData bp free Byte2dImageData bp free bp im_end_func byte_2d_destroy_image return New image types 12 5 Image 2 1 Library User manual Copying a part of the image Regions Of Interest ROI are handled entirely by the infrastructure An image processing operation does not have to perform any special action to process a region of interest as long as it calls pre_op before it starts to process the data The infrastructure pre_op copies the data from the parent of the ROI into a separate image The ROI image can then be processed by the operation Finally if the output image is also a ROI post_op copies the data from that ROI image back into its parent For copying the image data from the parent to the ROI and back to
93. Nee ee ene Store in file If you choose PERI and CR as choices for the shape argument the appropriate measure function is called with the value 214 6 filename is used when the argument is the name of a file For this argument type the min and max fields are used to supply the SCIL_Image file selector with necessary information The min field is used to specify a default directory The max field may contain several patterns to be matched in the file name You can use this for instance to specify the suffixes of the filenames you want shown in the dialog box like im if you only want to see image files Only the files in the folder which match the patterns in the max field will be offered for selection An example image cdf UNC readfile ENU I O ARGS filename con trui x ics im dat tif Filename image out A USE_NAME 0L Image int con 60 1000 X Position int con 60 1000 Y Position Advanced SCIL_Image 5 13 SCIL_Image 1 4 User Manual filesave is identical to filename on X window systems On other platforms the filename type may not always offer an editable text field to fill in a new name for a file while a filesave type will image cdf UNC writefile ANUAL writefil ENU I O I_O ARGS image in A Image filesave con ast TAN Filename choice con ICS_F ICS_F 1 TIFF_F 2 TCL_F 3 JPEG F 4
94. OF rmvar load chain since they always clear SCIL_Image s memory 4 8 The C Interpreter SCIL_Image 1 4 User Manual This does not mean that you have to include all your functions in every new program that you write since otherwise they might not be known because SCIL_Image offers the UFO This means Undefined Function Obtainer and describes SCIL_Image s capability to find non compiled user functions The functions not defined in a program are automatically loaded into SCIL_Image s memory provided that e Each undefined function is defined in a file with the same name followed by the suffix 2 x C e Each file can be found in the current directory or in one of the directories in the environment variable MACRO Such a function is retrieved automatically and executed Like your program the UFO will stay in SCIL_Image s memory until you give aload rmvar or chain command When a list command is given both the text of the loaded program and the text of the UFOs are shown Program Development Commands In this section we give the commands which can be entered from the Direct Command Mode and which are used for program development Some of these commands have been discussed before others are new Here is a summary chain lt file gt args load and run program list start end show program text load file load program text logon lt file gt connect logfile to session logoff disconnect logfile
95. Scilinit L Prog image2_1 txt brochure pdf a brochure about SCIL_Image comfile SCIL_Image command description file grab32 dll sample framegrabber dll grabcol dll sample framegrabber dll image2_1 txt ASCII tekst file with additional notes license scl file which must contain your license code readme doc release notes regarding this version of SCIL_Image scilimag exe the SCIL_Image executable 2 The hierarchy of the files and directories in SCIL_Image was carefully planned so please do not move files or directories to other directories before you know exactly what the files directories are used for If you do move them be sure that SCIL_Image can locate them see Section 2 3 2 4 Setting up SCIL_Image SCIL_Image 1 4 User Manual scilinit the initialization file for SCIL_Image It contains the following directories The demo directory This folder contains demonstration files with examples The help directory This folder contains SCIL_Image s on line Help facility The images directory This folder contains images Most of these images are used by the demos so do not remove them in order to save space The prog directory The prog directory is forSCIL_Image support and contains files and programs which you need to build your own version of SCIL_Image We will explain the procedure for that in Chapter 5 comfiles Esl comfiles mak overload obj example j
96. TIONS NOT_CO ARGS PILE D filenam trui My File image A My Image odd 127 1 255 Only odd values Before we begin describing the syntax we have a warning IMPORTANT MAINTAIN BACKUPS Please take extreme care when you make changes to your CDF If you make a mistake in your CDF this mistake propagates into the comfile and SCIL_Image will not start at all Be sure to keep a backup of the last working version so you can restore the old version if needed The following sections are written in a formal syntax to prevent ambiguities The syntax of the Command Description File is Advanced SCIL_Image 5 3 SCIL_Image 1 4 User Manual CDF lt CDF_entry gt lt CDF_entry gt lt comment entry gt lt menu entry gt lt translate entry gt lt variable entry gt lt command entry gt A CDF may consist of one or more CDF entries the means one or more of the previous A CDF entry can either be a comment entry or a menu entry or a translate entry or a variable entry or a command entry comment entry A lt comment entry gt is indicated by a hash sign at the beginning of a line lt commententry gt lt any text gt menu entry The lt menu entry gt is used to specify the way the menu looks and the placement of your commands in it The syntax of a menu entry is lt menu entry gt i lt menu_name gt
97. The histogram objects in SCIL_Image behave very much like images When you choose a existing histogram object to hold the histogram data of an image the histogram object will automatically be reconfigured For instance if we calculate the histogram of a color image the histograms for the red green and blue channel are shown separately readfile flamingo A show_histogram A The histogram of any image type can be shown except of complex images in any number of bins Also it is not necessary that all pixel values in the image are put in the histogram Consider this example a floating point image containing values of between 1 0 and 1 0 of which only the data between 0 0 and 1 0 is shown in 100 bins convert A B float_2d sin im BB show_histogram B 100 0 0 1 0 To inspect the contents of the histogram a list of all values can be dumped to the terminal of stored in an ASCII file using the dump_histogram command dump_histogram histogram_of_B 10 The command show_histogram_info gives some additional information about the histogram object like number of bins width of the bins and median value of the lowest and highest bin The Image 2 1 library in SCIL_Image 6 27 Chapter 7 Introduction to Image 2 1 This chapter introduces the Image library Read this chapter if e You want an overview on processing images with Image e You have basic knowledge of image processing Knowledge of the C language is not required for this cha
98. User Manual The Programming Mode Interpreted C functions and UFOs To use SCIL_Image in the Programming Mode rather than the Direct Command Mode means that you use an editor to create or change a program after which you load the program into SCIL_Image and run it This process may be repeated several times just as in ordinary programming The basic development scheme is edit load run edit etc The program may consist of includes defines global variables and functions with local variables just as in standard C This allows you to define C programs for the SCIL_Image C interpreter There is no need to compile them To make an interpreted so non compiled C program do the following 1 Create a file using the editor You may use SCIL_Image commands but we advise against this since you would have to redo those commands if you were to decide later that you wanted to make the functions in the program into compiled functions see Advanced SCIL_Image on page 5 1 Please note that almost all standard C syntax is permitted the only exceptions can be found in Features of SCIL_Image s C interpreter on page 4 16 2 Save the file in a directory listed in the environment variable MACRO Chapter 2 You should choose a name with the format lt name gt c to distinguish the program from a macro file 3 Run the program There are three ways to run it e From the menu by select File Run only if your program has
99. User manual image h Since nothing can fail while when adding all checking has been done in g_add and overflow is unlikely since each pixel is 16 bits we do not return a status Image supplies several example files to illustrate the mechanism of overloading These files can be found in the src_exmp directory In that location directories are located for a few image types In each of these directories source files for the corresponding image types are located In the grey_2d directory the files arith c and arithlow c contain the functions g_add and 1_g_add Compare these functions with the functions f_add and 1_f add float_2d directory to see the difference in how grey valued images and floating point images are handled Further several examples of generic level functions are found in the generic directory Overload tables Image determines which function it must call to perform a certain operation on an image type by means of overload tables The programmer specifies the type specific functions in a plain text file the overload files These files can be recognized by the ovl file extension At compile time a special utility program mkoverld converts these overload files to a C source code file overload c containing several arrays This C file is then compiled and linked with the Image library The layout of the overload files is not very complicated and looks like this optional comment
100. You might as well destroy the extra viewport on image A since we will not be needing it any more double click on the close box of the extra viewport A click Viewport only in the dialog box Zoom and pan can also be used to view color images or 3 dimensional images select Image I O readfile click Browse double click c scilimage14 images chromo3d ids click OK This is a 3 dimensional image of a cell division When you select it with the right mouse button and activate zoom and pan you will see that its dialog box contains buttons for z z and a scroll bar These can be used to step through the stack of 2 dimensional images that form the 3 dimensional image t g3D 1607140716 12 f Zoom amp Pan am 2s o BE F Help z Center A Figure 3 8 Viewing Z slices A utility that differs from zoom and pan in a useful way is lens The lens command allows you to construct an image that contains only part of image A instead of just a displayed viewport Let us read in a grey valued image 3 10 Getting started SCIL_Image 1 4 User Manual select Image I O readfile click OK select ITools lens select Lens image B Image B becomes a 31 31 image since those are the default values for lens To view it properly select it with the right mouse button click on zoom and pan make the zoom factor 8 then close the zoom and pan Now you c
101. _Image 1 4 User Manual select Image I_O readfile The menu will disappear and a SCIL_Image dialog box will pop up Gare OX Filename tu Browse Image c E X Position 300 Y Position 300 Doit Dismiss Heip _ Figure 3 3 A dialog box A SCIL_Image dialog box may contain several items The command you just selected is the title of the box in this case readfile The four buttons at the bottom are found in all dialog boxes button executes the command and removes the dialog box button executes the command but keeps the dialog box button gives on line help information about the command button closes the dialog box it has the same effect as clicking the close box in the upper left corner of the dialog box There is no need to make the dialog box disappear before you open up another one SCIL_Image will retain the last one if they conflict The default image trui is not very interesting you may view it by clicking OK To select an image from the library that is included with SCIL_Image select the browse button click Browse A dialog box Open appears in which you can select an image file in the normal way you select files in Windows The images are in the sub directory images of the directory scilimag You will find a number of files Let us select schema dat by double clicking Getting started 3 5 SCIL_Image 1 4 User Manual
102. _STATUS estack estack get_im_error_stack spb_subscribe estack other parameters For further information regarding this error stack please refer to the chapter Programming with Image Publish and Subscribe 8 9 Chapter 9 Programming with Image This chapter describes how to write image processing functions for use in Image Do not read this chapter if e You are a novice user e You want to use existing routines only e You have no experience with programming in C Read this chapter if e You are an experienced user e You want to write your own image processing routines e You want to know more about the image processing infrastructure Image 2 1 Library User manual Introduction to Image This chapter is on developing image processing routines and incorporating them into an image processing application that uses the Image Library A solid knowledge of the C language is assumed This chapter contains the information needed to add image processing functions to the library or to modify existing operations For a developer an important feature of Image is the support of different image types Knowledge of the infrastructure is therefore essential for creating functions that make maximal use of the library An overview is presented in Image infrastructure on page 9 3 A typical routine in the Image context reads as follows function heading images parameters declaration of an image d
103. _get_image_window_info IMAGE im char buf im_begin_func g_2d_get_image_window_info sprintf buf byte2D d sd ImageWidth im ImageHeight im im_end_func g_2d_get_image_window_info return Conversion to other image types By implementing two functions per image type one to put the data in an intermediate data space and one to retrieve the data from that space the data of any image type can be converted into any other image type The intermediate data space is described by a COMMON_LINE structure and can hold one line of image data multi channel is possible This data is stored in either long integer format or in double precision floating point The functions that puts the data in the COMMON_LINE conv_to_common determines which data format is chosen allocates the necessary data space and fills the COMMON_LINE structure to describe the data The function of the receiving image type conv_from_common must be prepared to handle both the integer and the floating point data These functions are repeatedly called by the convert function until all lines in the image are converted one by one See Data conversion convert in chapter Programming with Image for more information The function conv_to_common should return OK 1 as long as not all lines have been written to the COMMON_LINE and NOT_OK 0 when it has written the last line Its header void type_conv_to_common IMAGE im int n
104. _im_display_buf 0 1 In the occasion that the window containing the information is unwanted auto_point No If this command is used no window will pop up To restore the data display auto_point Yes Displaying the image At the completion of each operation on an image the output image is displayed If automatic image display is unwanted it can be disables with the command auto_display off From then on output images will not be displayed For example copy_im A C Operations are considerably faster with the display off To turn back on auto_display on copy_im C D The image in D is now displayed but image C is still invisible To display image C display_image C Both image C and image D are now visible in the display windows The set_display_mode command enhances the visual contrast of an image without modifying the image data Try The Image 2 1 library in SCIL_Image 6 5 SCIL_Image 1 4 User Manual set_display_mode C LIN_STRETCH For some images the difference is marginal as is seen in the displays of images A and C The pixels in image A already covered nearly the entire range of values that the display can handle So in this case there is little improvement in contrast Displaying 3D images In SCIL_Image 3D images are displayed one slice at the time To see how this works read in a 3D image readfile chromo3d C To display a different slice there are three options One is to click the rig
105. _info As with implementing a new image type the last three functions from the list above are only required when defining an image subtype in the complete SCIL_Image environment The dedicated service functions and the overload file must be added to Image as described in previous sections The image subtype is then implemented and can be used Further image processing operations can be added either by inheritance from the main type or by adding a new operation Inheriting an operation from the main type is done by copying the entry in the overload table from the main type to the overload table of the subtype New image types 12 21 Chapter 13 Creating an Image Application This chapter discusses the topic of creating a stand alone application using the Image library Read this chapter if e you want to build an application with the Image Library Image 2 1 Library User manual Introduction The Image Library can be used to build your own image processing application The library can best be regarded as an application without a front end Whenever a function is called in the library the behavior and state of entire library may be affected by it To use the Image library only one obligatory step has to be taken That is e Initialization of the library before any image processing function is called See Initialization on page 13 2 Once this requirement is satisfied the conditions are set for proper
106. a Makefile D scilres include makefile lib scilimagel 4 dsp lib i Makefile scl scilimagel 4 dsw mylib P mkoverld exe c sysfunc c overload E mksysfne exe sysfunc obj resource e overload c J stc_exmp i overload mak comfiles directory with Command Description Files example directory containing some of the examples used in this manual icons directory with the icons used by the SCIL_Image user interface include directory containing include files lib directory containing the SCIL_Image functions mylib directory to store yourown SCIL_Image functions overload directory with files for command overloading resource directory with resource files and icons src_exmp directory with sample C files referred to in Chapter 8 comfiles mak makefile lib scl mkoverld exe external makefile for Microsoft Visual C sub makefiles used by makefile program to create the file overload c mksysfnc exe program to create sysfunc c from the comfile overload c c file needed to create a new SCIL_Image executable overload mak external makefile for Microsoft Visual C Setting up SCIL_Image 2 5 SCIL_Image 1 4 User Manual overload obj compiled overload file scil res compiled resources scilimage14 dsp project file for Microsoft Visual C 5 0 scilimage14 dsw project workspace file for Microsoft Visual C 5 0 sysfunc c c file needed to create a new SCIL_Image executable sysfunc obj compiled sysfunc file Your SCIL_Ima
107. a 256 256 grey value image Dv be 8 26 x 11 69 in Figure 3 28 a reference manual page 3 44 Getting started Chapter 4 The C Interpreter When you write programs in a high level programming language like C you have to follow the sequence edit compile load run With SCIL_Image it is possible to just edit and run your program The part of SCIL_Image that makes this possible is called the C interpreter In this chapter we discuss the use of this interpreter for developing software You can compile and load if you want to make your programs run faster consult Advanced SCIL_Image on page 5 1 Read this chapter if You have experience with the C programming language You are experienced with the menu and command modes You want information on the use of macros You want to combine image processing commands with C control statements You want to develop image processing programs You need to know the limitations of SCIL_Image s C language Do not read this chapter if you have no experience with SCIL_Image you have never programmed in C you only want to use SCIL_Image at the Menu Mode of interaction SCIL_Image 1 4 User Manual SCIL_Image and C After starting SCIL_Image you can enter any SCIL_Image command or C statement into the Worksheet It will be interpreted immediately after you hit the Enter key nterpreted means that SCIL_Image e Identifies the C code that will execute your comma
108. a GREY_2D or a GREY_3D image Then the image out is changed into the type and sizes of image in If nothing went wrong pre_op will return OK 1 and processing of the data can commence Output of specific type If the output must be a specific type the last parameter is used to specify the type To threshold a GREY_2D image the output image must be converted to BINARY_2D To accomplish this pre_op in out ADJUST G_2D_SPEC BINARY_2D Although the low level routine in this case can handle both 2d and 3d this call to pre_op can only allow GREY_2D image for input This is because the last parameter of pre_op can only hold one image type at a time The calls topre_op for 2d and 3d images must therefore be separate calls when the output image must be of a specific type Only output Some operations only have an output image It may be that the output image is required to be of a specific type The pre_op routine should then accept all types of image as input and convert the output image to the desired type Assume the output image should be of type 9 16 Programming with Image Image 2 1 Library User manual GREY_2D then either of the following calls to pre_op will achieve the desired conversion pre_op out out ADJUST WHATEVER GREY_2D pre_op out out ADJUST ImageTypeSpec out GREY_2D WHATEVER stands for all possible type specs and ImageTypeSpec out is the sp
109. a is multi channel data like RGB color images the histogram can be stored in three channels to get an overview of how each of the three colors are distributed e nr_dims is the number of dimensions of the histogram The histogram objects can handle up to V_ O_MAX_DIM 5 dimensions e dimisan array of V_O_MAX_DIM 5 unsigned int s in which the dimensions are stored e name is a array of IM_NAMELEN 80 char s in which the name of the histogram objects is stored This name is used by the dialog boxes to identify the histogram objects e comment is a pointer to a string of arbitrary length A comment string of any length can be attached to the structure using the histogram_comment function This function allocates memory for the string and frees the memory of an previously added comment string e lbin_median is a double that describes the center pixel value of the lowest cell of each dimension in the histogram see example below e hbin_median is a double that describes the center pixel value of the highest cell in the histogram see figure below e bin_width is a double that describes the width of the cells of the histogram see figure below lowest bin highest bin LBinMedianValue N BinWidth HBinMediaaValue Programming with Image 9 41 Image 2 1 Library User manual The fields Lbin_median hbin_median and bin_width describe which pixel value of the image is stored in which cell of the histogra
110. a set of image processing commands or a tailored program made available to the user usually in the form of a menu option It is possible that the standard image processing operation of SCIL_Image are still available to the user but sometimes they are not This is not the choice of the SCIL_Image team but rather a choice of the developer of that specific application Using the system at the application level boils down to selecting options with the mouse and filling in parameter values where needed The manual of the specific application is not a part of this manual as this manual describes only the generic use of the SCIL_Image system To get an overview of the standard SCIL_Image system the reader is referred to the chapter Getting started on page 3 3 1 for introductory sample sessions The remainder of this manual is about using SCIL_Image for interactively analyzing images using the extensive image processing library of SCIL_Image for building image processing applications Using the system as an interactive user At the highest level of use functions can be executed one by one by interactively selecting them from the menu system All operations of SCIL_Image are available to the interactive user ranging from reading an image displaying the image to enhancing the contents making a histogram or a Fourier spectrum performing mathematical morphology and so on A quick overview of the functions is given in the guided tour through the system
111. acteristics of one image type Also not having to change existing source code when implementing an operation for another new image type has considerable advantages This is explained in depth in Function overloading The Image types In Image of each basic image type a 2 dimensional and a 3 dimensional variant is present For each of these image types the pixels are stored consecutively in memory First the pixel of the first row then the pixels of the second row etc The 3D image types are laid out in memory as a set of 2D images first all the pixels of the first Z slice then all the pixels of the second Z slice etc Grey valued images Grey valued images consist of pixels each of which contain an integer value These pixels are represented in Image by a in the C language defined datatype PIXEL This type is defined tobea short int which means that on most computers 16 bits 2 bytes per pixel are used Therefore the value ranges from 32768 to 32767 a total of 65536 values Binary bitmapped images Each pixel in a binary images has value 0 or 1 One bit is therefore sufficient for the storage of each pixel In Image we store thirty two binary pixels in each 32bit integer long word so that a binary image requires only a fraction of the space occupied by a grey valued image A border of long words is reserved around each binary image to enable very fast morphological operations Floating point image
112. actically correct In this case it actually declares the variablei despite the incorrect syntax You can see this by trying it again now with the semicolon int i int i gt variable redeclaration According to the 1978 Kernighan and Ritchie C standard K amp R C C compilers will not generate an error or a warning if a function declaration and a call to that function have a different number of arguments This can lead to bugs which are extremely difficult to find To avoid these problems SCIL_Image checks on the number of arguments in function declarations and function calls and generates an error message when they differ As an example the command stremp which calls the function st rcmp expects two arguments i strcemp Just one argument i stremp Just one argument gt more function arguments expected i stremp Here are three arguments me rre i stremp Here are three arguments gt too many function arguments The C Interpreter 4 15 SCIL_Image 1 4 User Manual A K amp R C compiler would not have complained in this case SCIL_Image thus checks syntax more strictly than you may be used to When you explicitly test for errors for instance in opening a file it is common practice in C programming to leave the program with exit upon an error However you should be careful doing this because it will actually quit SCIL_Image rather than just the program which may not b
113. ae s S araia coat E aaa a i wade 13 3 Colon Eook ptabl si iini e a Galt E E E eda eter die 13 3 Sample Codes a RAG R eae eee A ed ese dee Mahe ade eee 13 3 Chapter 1 Outlines of SCIL_Image In this chapter an outline of SCIL_Image is given Because SCIL_Image is a multi layered and multi purpose system this may help you in orienting yourself in the system s use Read this chapter if e You are using the system as an interested user of image processing applications e You are using the system to develop image processing routines or applications or e You are to install the system or e You want to understand the SCIL_Image philosophy Do not read this chapter if e You are using an application based on SCIL_Image and SCIL_Image and the application have been properly installed for you Read the application manual and in a later stage read this manual if necessary SCIL_Image 1 4 User Manual The four layers of SCIL_Image SCIL_Image is an extensive multiple layered system for image processing and for the development of applications in the image processing domain As a consequence of the layered structure SCIL_Image can be operated by users with various levels of expertise with image processing and with various levels of programming skills from a user of an image processing application to an image processing system developer Using the system at the application level An image processing application is
114. age 1 4 User Manual We wrote such a function to implement the image type as a variation of the choice type It does the following 1 Ifthe type is not image then 0 is returned 2 The type is changed to choice 3 minstr which is the variable corresponding to the min field is changed to hold a list of all the available image names with the image address as an alias So if the images A B and scratch are available nstr will be changed to A 234568 B 236888 scratch 238444 or something similar depending on the image pointers 4 maxstr which is the variable corresponding to the max field may be used to pass on extra information such as a symbolic constant Therefore if the maxst r field is not equal to the minus sign the information in the field is added to the list with choices in minstr 5 The value is returned to indicate that our function has successfully transformed an unknown type into a known type in this case image to choice Several types have been added in this way These types are image im_type v_object histogram clut presents a list with available images presents a list of image types used for instance in the Image Conversion convert command presents a list with available var_objects see Non image data var_objects page 6 24 The min field can be used to specify the var_object class If this field is used only var_objects
115. age data we encourage you to use these whenever possible Although we refer to most messages by name you must remember that all messages are numbers that must be unique for that event message One message deserves special attention the SPB_DESTROY message This message signals the destruction of an object Because the information regarding subscribers is stored within the object itself the publish field the destroy message must be published before the object is actually destroyed Another good reason for publishing the destroy message beforehand is the possibility that subscribers may still need to access the object Whenever a subscriber receives the SPB_DESTROY message it knows that this is the final message from that publisher The subscriber can not prevent the destruction of the publishing object It must clean up everything that relates to that object and may even decide to destroy itself It is not necessary to unsubscribe from an object that is destroyed the mechanism will automatically unsubscribe all subscribers when an object publishes a SPB_ DESTROY message Finally the message values 0 to 9999 as well as the message prefixes SPB_ IMP_ and AF_ are reserved If you wish to define additional messages for the use in an interface or elsewhere use values outside the above mentioned range and symbolic names that do not use the above mentioned prefixes Publish and Subscribe in the Image library The Image libr
116. age have no means to visualize error information By using the error mechanism 9 22 Programming with Image Image 2 1 Library User manual functions in Image supply the user interface with as much information about the error as possible When an error is detected it is reported to the im_report_error function This initializes the error stack IM_ERROR_STATUS struct from im_error h by copying the name stack to it and setting the error value for each function in the stack to TIE_NOT_OK 0 The function that reports the error also supplies an error value and a message which are both stored in the stack The function then returns with an error status and its caller further handles the error situation In its turn it will clean up and add its own error status and message to the error stack with the im_report_error function and also returns In this manner all the functions in the error stack handle the error and return Because the error mechanism knows the exact location of the error only the relevant functions can add their error values and messages to the stack If during the clean up phase of a function another error occurs and a function that is not in the error stack reports this the information regarding it will be ignored by the error mechanism When the top level function of the error stack is reached the error mechanism will publish that an error occurred For an application this means that two ways a
117. age itself This publish is followed by a SPB_CHANGED publish of this image because the contents changed as well SPB_MOVED is published by a ROI only it signals that its position inside the parent changed The data is a NULL pointer the new location can be retrieved from the image itself SPB_RESIZED is published when the sizes of the image changed The data is a NULL pointer the new sizes of the image can be retrieved from the image itself This publish is followed by a SPB_CHANGED publish of this image because the contents changed as well SPB_CREATE_ DISPLAY is published as a request to the display interface if present to create a display window for this image The data is a spbAREA3D structure containing the desired location and sizes of the display window Color lookup table publishes In Image 2 1 a color lookup table CLUT may be an integral part of an image The definition of the CLUTs allows a display interface to store some window system dependent data in the CLUT To give a display interface the opportunity to react to changes to the CLUTs these Publish and Subscribe 8 7 Image 2 1 Library User manual changes are published via a top level CLUT This top level CLUT is called super_clut and is globally available as a void pointer or via the function get_super_clut void s_clut s_clut get_super_clut spb_subscribe s_clut NULL funcname NULL It currently publishes the following messa
118. alue that is present in image C has to be determined This can be done by 6 14 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual pix_maxval C In the command window a line is printed that states maximum 0 934 Now image C can be stretched by multiplying all values by 255 to stay within the range 0 255 eval d c 255 Complex images COMPLEX_2D amp COMPLEX_3D Data representation of complex images Each pixel in a complex image is a complex number and therefore has a real and imaginary part In SCIL_Image each of the parts is represented with the C type float Try the following readfile trui A eval B 0 333 A complex_im A B C Image C now contains a complex image each pixel of which has its real part from the corresponding pixel in image A and its imaginary part from the corresponding pixel in image B Examining the pixels in image C with the left mouse button results in the following text XXX YYY Re VV VVV Im VV VVV The first value after the parenthesis is the real part of the pixel and the second value is the imaginary part Usage of complex images This type of image is specifically required for the Fast Fourier Transforms FFT Examples of operations on complex images To calculate the FFT fast_fourier C C Forward Basic arithmetic operations such as adding and multiplying can be applied to complex images or the real and imaginary parts can be examined an
119. alues can now be processed by other routines or be saved for later reference Storage of var_objects on disk Like images var_objects can be stored on disk for later use The format in which the data is stored resembles the ICS format Again two files are used to store the object and its ASCII header The data file carries the extension vod Var_Object Data and the header file the extension voh Var_Object Header The header is ASCII to enable easy reading and the comment attached to a var_object is also stored in that file Histogram objects The histogram of an image or a part of an image can provide valuable information on the image The information can be used for contrast enhancement segmentation statistical analysis and for normalization of the grey tone deformations through the equalization command SCIL_Image stores the histograms it calculates in so called histogram objects The histogram objects can be of any size and multi dimensional up to 5 dimensions Stored in the object is not only the histogram data itself but also the number of bins the width of a 6 26 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual single bin and the median value of the lowest bin and the highest bin all this for each dimension of the histogram The histogram can be made visible in separate window that can be resized and moved around by the show_histogram command readfile trui A show_histogram A
120. amain e Type in the Worksheet load lt name gt c run e Type in the Worksheet chain lt name c gt These three commands load run and chain are also available under the SCIL menu We demonstrate this by an example Type it and save it in a file which you name simpprog c The C Interpreter 4 7 SCIL_Image 1 4 User Manual simpprog c define STRINGSIZE 80 char string STRINGSIZE oo main int i 0 for i lt 2 1i printf Enter a string gets string printf The number of characters is d n charcount string int charcount s char s int cnt 0 while st 0 cntt return cnt Now tun it select File Run Enter a string WE NOW TYPE A STRING The number of characters is 20 Enter a string THAT WAS IT The number of characters is 13 After a program is loaded any of its functions may be called from the command line For example load simpprog c printf d n charcount How many characters am I 24 However during a SCIL_Image session you may want to load and run a new program This will delete all definitions made in your old program of variables and functions from SCIL_Image s working memory The two commands that load a new program and cause this loss of memory are load and chain Sometimes you want to clear the working memory the command rmvar does this But remember BEWARE
121. an move your mouse through image A and see the data in image B change along with it The other buttons on the status bar contain interactive tools that can be used for image display and also for simple image processing They deal with grey values and are therefore not applicable to all image types The interaction tools can also be used in combination For instance we can have the histogram of image A even when we are using the lens select image A by a left mouse click click the Histogram icon You can grab the green bar to read grey value and frequency from the histogram We can even have a histogram of the lens image B at the same time updated while we are moving the lens select image B by left mouse click click the Histogram button move around in image A by dragging the lens The histogram changes as the image in B changes because the lens position changes When you are done with the lens you have to switch it off explicitly using the menu select ITools stop_lens Another histogram based utility is the Threshold Editor which allows you to select an appropriate threshold for an image by looking at its histogram select image A click the Threshold Edit icon Getting started 3 11 SCIL_Image 1 4 User Manual The histogram of A is displayed and by dragging the green bars you can see the effect of a thresholding on A you can move both bars at the same time by dragging in the middle experiment with Th
122. an normal IMAGE structure to which a small structure is added An image is a ROI if the field roi in the IMAGE structure points to a ROI structure as defined in the include file roi h typedef struct roi_t IMAGE parent BOOL_MASK bool_mask int startx int starty int startz ROI The information describing a ROI is scattered over several structures an IMAGE structure a ROI structure and the structures which also are used with normal images Looking at the ROI structure it can be seen that little additional information is needed to describe a rectangular or cubic ROI Just the image in which the ROI resides parent and the top left corner of the ROI in the parent image start x starty startz All other information is in the IMAGE structure itself The field BOOL_MASK is used for defining non rectangular ROIs and is discussed in Region of interest data structure arbitrary shaped on page 9 9 When a ROI is specified as the input or output image for an operation the region is copied to a separate image by the pre_op function After the operation post_op copies the output result back into the parent image see Dynamic adjustment Pre_op Post_op on page 9 12 9 8 Programming with Image Image 2 1 Library User manual Region of interest data structure arbitrary shaped In addition to rectangular ROIs support is provided for the creation of arbitrary shaped ROB The BOOL_MASK fie
123. ary uses the publish and subscribe mechanism for several types of objects The images the histogram objects and the error object all publish events that happen When you wish to subscribe to image and other objects a problem arises To subscribe to an image you must have a pointer to that image But when you write your source code this pointer is not yet present Image solves this problem by supplying an administrative top level image that is permanently available This top level image publishes the creation of images Top level publishes As mentioned above to obtain the pointers to images you wish to subscribe to you must first subscribe to the top level image This top level image is called super_im and is globally available as a void pointer or via the function get_super_im Publish and Subscribe 8 5 Image 2 1 Library User manual void s_im s_im get_super_im spb_subscribe s_im NULL funcname NULL The pointer to this super_im can also be retrieved with the get_super_im function In the example above the function funcname receives a notification whenever new images are created The function receives as its third parameter a message ID in this case SPB_ITEM_ADD see also Receiving messages on page 8 3 Its fourth parameter is a data pointer which is the pointer to the new image The message IDs are defined in the include file spublish h Below a complete overview is given of all current mes
124. as handled an error by cleaning up and returning with a fault status the application can then review the information Text output handling The default behavior of the Image library is to show all text output through the stdout and stderr streams An application can overrule this by supplying a function pointer to which the text must be redirected see Textual output on page 9 26 Color Lookup tables A principal feature of the user interface of an image processing application is the display of the images Often color lookup tables are used to influence the representation of the images on the screen either to enhance the contrast or highlight specific features in the image Because the Image Library already supplies color lookup tables it is a good idea for the display interface to also make use of these tables instead of defining a new set The Image Library defines that when the contents of a clut is changed by an operation this must be published through the super_clut object The display interface is then able to react to the changed CLUT providing it subscribed to the super_clut object For additional information on cluts see Image Color Lookup Tables on page 9 11 Sample code The Image library is accompanied by a sample program to show some of the important issues involved in using the library In the listing below we have numbered these issues and they are discussed in more detail The actual image
125. ata structure The IMAGE structure calls to parameter checking routines Checking routines call to pre_op Dynamic adjustment Pre_op Post_op processing A simple example error handling Error handling and reporting call to post_op Dynamic adjustment Pre_op Post_op function ending The IMAGE data structure is discussed in The IMAGE structure on page 9 6 In Dynamic adjustment Pre_op Post_op on page 9 12 the obligatory pre_op and post_op routines are considered as well as the dynamic size and type adjustment facilities An example of an image processing function is found in A simple example on page 9 19 Routines for checking non image parameters are described in Checking routines on page 9 23 Error handling and reporting on page 9 21 is devoted to error handling For textual output a special function is defined intended to replace the default print f and fprintf functions thus enabling a User interface to influence the presentation of all text output The section Textual output on page 9 26 discusses this subject The function overloading mechanism described in Function overloading on page 9 27 is a useful tool for keeping functions simple which must implemented for multiple image types In Data conversion convert on page 9 32 the design of routines which involve a conversion of the image type is considered Whe
126. bar of the display window confirms that this is a binary image b2d threshold mapped all pixels in image A with value less than 128 to 0 pixels in image B and pixels that had a value of 128 or more to 1 pixels The colors in which the binary images are displayed can be changed by using the bin_disp_colors command bin_disp_colors WHITE BLACK display_image B Usage of binary images Because a binary image is made up of pixels each of which is restricted to one of two values the only information contained in a binary image is limited to which pixels are associated with an object and which are associated with the background Binary images therefore play an important role in the investigation of object shape and form Examples of binary operations Noise can be filtered from binary images as follows percentile B C 334 To extract object skeletons in an image readfile cermet A threshold A B 128 invert_im B B hild_skelet B C Often only objects with a pre defined minimum size are of interest Removing smaller object can be accomplished as follows readfile cermet A threshold A B 128 invert_im B B The Image 2 1 library in SCIL_Image 6 13 SCIL_Image 1 4 User Manual erosion3x3 B C 8 propagation C B D 308 By eroding several times no seed is left by small objects for thepropagation operation that enlarges the eroded objects to their original sizes Floating point images FLOAT_2D amp FLOAT_3D
127. ber of commands Their effect in all windows is the same Enter on main keyboard gives a new line Enter on numeric keypad or Shift Enter on main keyboard executes a selection and reports that execution in the Worksheet and the History window If no command has been selected the line on which the cursor presently resides is executed Ctrl Enter pops up a dialog box for the selection By clicking Help in that box you obtain on line Help information You will find these special key sequences very convenient when developing your applications 3 42 Getting started SCIL_Image 1 4 User Manual On line manuals The documentation of SCIL_Image 1 4 is supplied in Adobe PDF format To read these documents the Acrobat Reader supplied with SCIL_Image is needed The Help on SCIL_Image button from the Help menu produces a window as shown below Acrobat Reader scilimag pdf Mil El WIR File Edt View Tools Window Help 18 x D Table of conten D Chapter1 Outli D Chapter 2 Setti D Chapter 3 Getti D Chapter 4 The O Chapter 5 Adva O Chapter 6 The O Chapter 7 Intro D Chapter 8 Publ O Chapter 9 Prog D Chapter 10 Anz O Chapter 11 Bitr O Chapter 12 Nev D Chapter 13 Cre C Appendixes C Reference man g wep University of Amsterdam TNO Institute of Faculty of Mathematics Applied Physics Computer Science Physics and Astronomy VVVVVIVVVVVTVTF Kruislaan 403 P O Box 155 1098 SJ Amsterdam
128. bject contains it can have one of these values PIXEL_T CHAR_T SHORT_T INT_T LONG_T FLOAT_T DOUBLE_T e data is the pointer to the memory that holds the data stored in the var_object e nr_channel isa special dimension introduced to support data types with multiples of data such as like complex values more than one value is stored in each unit of the object e dimensions is the number of dimensions in the var_object It can range from 0 to V_O_MAX_DIM which is 5 at present This is equivalent to the number of dimensions in an array By using 0 dimensions a scalar can be stored in a var_object e dims is an array that contains the length in each dimension in the var_object e size is the total number of elements in the var_object It is the product of all the values in dims array The number of channels is not included in the product e bpelem stands for bytes per element and contains the number of bytes used in each element of the data The number of channels per element is not included If a var_object for instance has four channels nr_channel 4 and the type is char 1 byte then bpelem is also 1 e object_name is a string that contains the name of the var_object This name is used in the dialogue boxes and in the command window The maximum length of a var_object name is IM_NAMELEN see image h characters e object_class is the object class of the var_object belongs to In the comfile this name is
129. bjects Var_objects in SCIL_Image are objects that are used to store non image data Several measurement functions in SCIL_Image store their result in a var_object In fact a var_object is equivalent to an array in C It was introduced to prevent inexperienced users from having to declare and address arrays They are also convenient for performing checks on sizes and data types Behavior of var_objects A var_object is similar to an image in the sense that the size and type of a var_object are automatically adjusted if an operation returns a specific type and size A var_object can have a number of dimensions ranging from 0 to 5 so a variety of data can be stored in it from a scalar O dimensional to a multidimensional array maximum 5 dimensions Further a dimension called the nr_channel number of channels is added for compatibility with existing image types This is useful for instance for complex images where each pixel has two values associated with it Which var_objects are shown in the dialogue box can be controlled with a group name Comments of arbitrary length can be attached to help the user in remembering what is stored in the var_object Also the objects can be saved to disk By default four var_objects are available for the user called objl obj2 obj3 and obj4 An unlimited number of extra var_objects may be created Information about a var_object can be obtained with the function show_var_object_info obj
130. ble look action per image pixel This method will be denoted as the Straight Forward Implementation SFD We will also compare our algorithms with the queue based algorithms developed by Verwer and VanVliet VERWER These algorithms will be referred to as the VV V algorithms These algorithms process only those pixels in the image which need to be processed The first iteration of erosions and dilations using the VVV algorithm is slow because the queue containing the pixels to be processed needs to be build by scanning the entire image Subsequent iterations then are much faster proportional with the number of contour pixels Number of VVV SFI SIP Iterations ms ms ms ms able va b il ej al tbe l ab c 1 6 23 175 160 639 283 283 153 1216 2 11 43 187 171 723 583 516 273 2416 4 15 62 206 200 851 1150 1133 491 4900 8 19 82 237 255 945 2300 2366 858 9850 16 24 100 269 302 1089 4716 4800 1371 20066 32 28 121 288 305 1303 9600 7000 2155 41616 64 36 153 291 305 1414 9600 6983 3643 86016 128 44 187 291 305 1423 9750 7050 7178 151066 Table 11 2 4 Connected Dilation Shown are the execution times in ms for the 4 connected dilation for the 3 different original images denoted as a b and c see Figure 11 4 Table 11 2 shows the experimental results for the 4 connected dilation for several values of n where n is the number of ite
131. cess Negative error status on failure see im_error h SEE ALSO clp contrast stretch equalize tn_state threshold lookup Figure 3 15 Another manual page An alternative way to find this Help page is hit Ctrl H click the Search button 3 24 Getting started SCIL_Image 1 4 User Manual type the command name Proceeding with the example the last command for i 0 i lt 200 i i 10 thresh A B i is equivalent to for i 0 i lt 200 i i 10 threshold A B i Edit the former to produce the latter and execute it The whole line is now a proper C statement which does not need to go through command expansion to be executed since it contains an explicit call to the C function threshold Please be aware of the difference between C statements and non C statements If you want to know the proper C syntax for a command called lt command name gt call up its Help page by the procedure described above Note carefully that although it is possible to mix legal C syntax with non C commands in the SCIL interpreter it is strongly recommended to use only legal C syntax in your programs because they could then be compiled which will make them faster and added to a library so you can use them as functions in your programs Finally we show you how to collect a series of commands in a file which can be executed Such a file is called a macro Let us make a macro for the set of commands that gave us the contou
132. criber itself fun_ptr is a pointer to the function the subscriber used to receive the messages The subscriber then is disconnected from the object and it will no longer receive any messages from the object To uniquely identify itself to the mechanism it must provide exactly the same identifier and function pointer as when it subscribed itself to the object Receiving messages The function the subscriber uses to receive the messages from a publisher must have the following header void func void object void ident int mess void data void client_data object is the publishing object ident is an identifier for the subscriber Publish and Subscribe 8 3 Image 2 1 Library User manual mess is the numeric ID of the message being published data is a pointer to the message specific data if any client_data is a pointer to the data specified when subscribing to this object It is the responsibility of the subscriber to recognize the messages and act upon them correctly The subscriber must silently ignore unrecognized and uninteresting messages as these may be of interest to other subscribers to this object Publishing messages An object that wishes to publish messages and meets the structure requirement as mentioned in Object requirements on page 8 2 calls the function spb_publish spb_publish void object int message void data object is the publishing object itself message
133. d about the image First the name of the image is given For instance the names of the images supplied by SCIL_Image at start up are A B C and D This is followed by a code in parenthesis representing the type of the image see Table 6 1 for the meaning of the code Finally the dimensions of the image are given For A B C and D at startup this is 256 256 because each of the images contain 256 rows of 256 pixels For a 3D image a number indicating the plane or slice currently displayed is displayed in parenthesis following the other information The left mouse button If the left mouse button is pressed inside a display window information is supplied about the position and value of the pixel pointed at On UNIX and Macintosh a small window pops up 6 4 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual containing the information on MS Windows the information is maintained only in the status bar In GREY_2D images the information is shown in the format XXX YYY VVV Where XXX is the x position YYY the y position and VVV is the value of the pixel at that position If the mouse button remains pressed and the pointer is moved around in the image the information in will updated accordingly The small information box can be positioned in the title bar permanently not follow the cursor by use of the command point_im_display_buf 0 0 The default behavior can be restored again by point
134. d manipulated separately by The Image 2 1 library in SCIL_Image 6 15 SCIL_Image 1 4 User Manual storing them in different images They can then be rejoined in one image and further manipulated real_im CB mul_float B 1 34 B complex_im B C D fast_fourier D D Reverse Labeled images LABEL_2D amp LABEL_3D Data representation of labeled images Like a grey valued image a labeled image is made up of integers with the C defined datatype PIXEL The interpretation of the values in a labeled image is however entirely different Usage of labeled images The information in a labeled image is similar to that in a binary image The pixels in the image either belong to an object or to the background In a labeled image however each object in the image has its own number or label used to distinguish it from other object in the image The pixels in an image which have been determined to be part of an object and are connected are assigned a unique identifier Each pixel in a labeled image has a value identifying it as a member of a set or object Labeled images allow measurements such as size and shape to be performed on the image objects Examples of operations on labeled images A labeled image can be created by labeling a binary image readfile cermet A threshold A B 128 invert_im B B aio_label B C Image C is now a labeled image see the title bar When a labeled image is displayed in SCIL_Image differe
135. d pan button at the bottom of the screen We explained how to do this in the previous session select image B by a right mouse button click click on the zoom and pan icon use the slide bar in the dialog box to set the zoom size to 4 0 Now we do the fourier transform on the reduced image select Image Transform fast_fourier choose Input Image B choose Output Image C choose forward click OK Image C also appears small in its viewport You would like to have it zoomed in by a factor of 4 0 just like B You can do this by a single action click on viewport C with the right mouse button The zoomer instantly applies to C as well note the green border around C Getting started 3 15 SCIL_Image 1 4 User Manual ORIGINAL IMAGE FOURIER TRANSFORM B g2D 64764 BE EA E C c20 6464 BE Figure 3 10 The Fast Fourier Transform The header of window C reads C c2D 64 64 This c2D indicates that the window contains a 2 dimensional image of which the value at every pixel is a complex number You may check this by pointing at it the status bar at the bottom of the screen indicates the complex values SCIL_Image discriminates carefully among the various types of images Not all operations are implemented for all image types For example if you try to do a Gaussian filter on C select Filter gauss choose Input Image C choose Output Image D click OK
136. d the value of the pixels he points to with the mouse The auto point information is stored and returned in a string format The global string pix_val_buf is allocated by the infrastructure to store the string The pointer to pix_val_buf should be the return value of the function if all is OK Otherwise NULL should be returned The function header must be char type pix value_str IMAGE image int x int y int z The byte 2d auto point information is supplied by this function T char byte_2d_pix_value_str IMAGE ip int x int y int z BYTE_2D_IMAGE bp im_begin_func byte_2d_pix_value_str if range_ok x 0 ImageWidth ip X coordinate range_ok y 0 ImageHeight ip Y coordinate New image types 12 11 Image 2 1 Library User manual im_report_error byte_2d_pix_value_str IE_OUTRA return NULL bp BYTE_2D_IMAGE ImageIn ip sprintf pix_val_buf 3d Byte2dImageData bp long y Byte2dImageWidth bp x im_end_func byte_2d_pix_value_str return pix_val_buf The information to be displayed in the title bar of the window is supplied by a function that prints the type of the image and its sizes in an supplied string The name of the image is inserted by the infrastructure itself The function header void type_get_image_window_info IMAGE image char buf The byte 2d version void byte_2d
137. d your own functions in SCIL_Image in a way that makes them indistinguishable from the standard functions they are then indistinguishable from the standard SCIL_Image commands This means that you can make your own version of SCIL_Image completely tailored to your own application e You should use pre_op and post_op in your image processing programs to have the advantage of SCIL_Image s type checking error handling and automatic displaying of results e You can overload one command name to treat images of various types using the overload mechanism 3 36 Getting started SCIL_Image 1 4 User Manual This concludes the fifth session The Commands of the SCIL_Image Menus In this section we briefly describe the menus in the SCIL_Image menu bar This will give you an overview of SCIL_Image s capabilities We will specify the sections in which you can find more detailed descriptions of the commands File New C Program CTRL N Open C Program CTRL O Open Image 1p fh Save Image As Hur PE Print CTRL P Printer Setup Print Image Setup Exit SCIL OC scilimag PROGSEXAMPLE myfunc c Figure 3 18 the File menu This menu contains commands for file manipulation Most commands are standard SCIL_Image also provides the Open Image command to open any of the image files Edit Figure 3 19 the Edit menu This is the standard editor menu You can u
138. display function must be int type_part_image_display im sx sy sz width height depth IMAGE im int SX Sy sz width height depth The byte 2d example for sake of simplicity is only capable of size scaling and color lookup tables and does not consider the automatic display enhancements attached to the image For a complete description one should refer to the display sources as provided with the standard distribution It is also assumed that the standard configuration is an 8bit screen for 1bit and 24bit monitors again the reader is referred to the source examples nc nc nc ncl nc de lt stdio h gt de lt string h gt de lt stdlib h gt de image h de im_error h de im_infra h H H H H H H H H H H O H H 0 H H H H ncl ncl ncl ncl nel nel nel nel H n oO nel nel u u u u u lu u u u u u u u u u ndif de portab h de dmodes h de disp_im h de byte_2d h fdef MACINTOSH de lt Palettes h gt de Macviewport h de disp_mac h de xviewprt h de disp_x h New image types 12 9 Imag exte int IMAG int 8 int IMAG VIEW int e 2 1 Library User manual rn CLUT xstandard_cluts byte_2d_part_image_display im sx sy sz width height depth Bo im SX Sy Sz width height depth im_begin_func byte_2d_part_image_display if ImageVport
139. do not initialize them or operate on them in any way In most cases this requires that the service functions will have to be re implemented for the subtype The service functions are described in Implementing a new image type The defines of the type_ident the type_spec the type descriptor and the macros to access the fields of the descriptor should be put in a separate header file named after the image type In the case of the label_2d subtype the definition of thet ype_ident and the type_spec is located in image h and not in label_2d h As with a new image type an overload file must be created to tell the infrastructure which functions to call for this image type Looking at the overload file of the label_2d image type we can see that most of the service functions are inherited from the grey_2d image main type LAB EL 2D OVERLOAD TABLE ABLE NOTE LABEL 2D INHERITS MOSTLY FROM GREY_2D 12d LABEL_2D L_2D_SPEC 2 Basic functions for each image type create_image g_2d_create_image destroy_image g_2d_destroy_image copy_part_image g_copy_part_image copy_masked_part g_copy_masked_part conv_to_common g_2d_conv_to_common conv_from_common g_2d_conv_from_common 12 20 New image types Image 2 1 Library User manual part_image_display 1_2d_part_image_display pix_value_str g_2d_pix_value_str get_image_window_info 1_2d_get_image_window
140. ds e Ifyou want to execute more than one command in the Worksheet you select them by using the click and drag method until all desired commands are highlighted then hit the Enter key e Ifyou want to re execute a command you can do this simply by 1 moving the cursor to that line in the Worksheet 2 clicking to activate the line it will not look selected but it is 3 hitting Enter e You can also edit a command using the normal method mouse selection and re typing before you re execute it In the same way you can remove lines that you do not want Another way to re execute commands is provided by the History window In that window all commands you have given are saved in the order you gave them You can reach the History window under the Window menu F History Pel printf Hello world n readfile trui A reafdile trui A readfile trui A 300 300 readf set_window_size A100 120 int i for i 0 i lt 200 i i 10 thresh A B i for i 0 i lt 200 i i 10 thresh A B i Figure 3 14 The history window When you re executed the for loop it was listed in the history window a second time even though you did it by selecting the line in the Worksheet that was already there The History window is a faithful account of what you did It can not be edited You select lines in the same way as for the Worksheet and execute by hitting Enter SCIL_Image can pop up three different ki
141. ds by their name rather then selecting them from the menu In SCIL_Image a user may type the commands in the text window directly As all menu commands are translated into written commands typing commands may be mixed with selecting and executing items from the menus There is no difference Using the menu system is easier to learn and there is more support to the user in selecting parameter values and options but typing commands is quicker Therefore simultaneous use of the two interface levels combines the best of both worlds Control statements At a point connecting image processing commands may no longer be good enough to reach your image processing target Frequently streams of image processing commands are combined into a new high level image processing operation Or a set of commands is executed as many times as there are particles in the image Or some stream of image processing commands is repeated to improve the quality of the image until it cannot be improved any further Or an image processing measurement function results in a numerical value e g the image contrast or the number of particles in the image and that number is needed as a parameter value in another operation e g an operation to normalize the contrast or an operation to select all particles with a certain minimum area If such a control over the flow through the image processing commands is desired control statements are needed These statements cannot be
142. e dialog bokefore starting the measurements use the Browse button next to the bar to look through your directories You may need more image windows than the standard windows A B C and D for instance to store interesting results You can make a new image window by select Image make_image type as Image name my_image The image can have any name as long as it does not contain a white space type size and location you want by selecting the appropriate options in the dialog box Creating small sub images may be helpful in selecting details of larger images in which case you may find the copy_part_image command from the Image menu useful When you select a command Image copy_im for instance the choice of images includes your newly made image my_image You should realize that the image is lost when you exit SCIL_Image unless you save it using Image I_O writefile SCIL_Image will not prompt you to save the images you have to do it yourself This ends the second sample session To quit select File Exit SCIL 3 Clicking in the close box of the labeled image will destroy that image but not terminate the running program In that case you would still have to stop it by using the Pause Break key and Enter Your data is then not written to a file 3 20 Getting started SCIL_Image 1 4 User Manual Session Three Using the SCIL_Image Command Line Mode This sample
143. e Open as to text in the dialog box and at the end of the OVERLOAD list add c scilimage14 prog example add_one ovl and save the file fill in the appropriate path name if you saved the file add_one ovl somewhere else Getting started 3 33 SCIL_Image 1 4 User Manual 4 For both the comfiles and overload project choose the Build command on the context menu when clicking the project with the right mouse button 5 Then rebuild SCIL_Image by choosing the Build command on the scilimage14 project 6 Start SCIL_Image again In the Arithmetic menu you will find your command add_one Please note that just must build the comfiles and overload projects by hand first because Developer Studio does not recompile the files sysfunc c and overload c when necessary To test it do not read in an image just choose your new command either by typing or by selecting it from the Arithmetic menu After your command has been executed this is almost immediately you may check that it indeed added 1 Clicking with the cursor in image window A shows that it has value 0 and clicking in window B should show that it has value 1 Now convert image A into a real valued image by selecting Image Conversion convert and repeat your command Now it changes values of 0 0000 to 1 0000 Thus it works on real valued images as well Now that you know that the command works it is time to add a description of it
144. e and since C is an output image it is converted to GREY_2D with pre_op as described in Section Dynamic adjustment Pre_op Post_op on page 9 12 Programming with Image 9 27 Image 2 1 Library User manual Generic function layer The function specified as the generic function is called by the user application It calls the function overloader to acquire the appropriate function for the image type and then calls that function with its arguments For the add_im operation the generic function is necessary structures defines and prototypes include lt stdlib h gt include image h include im_error h include generic h include im_infra h int add_im IMAGE inl IMAGE in2 IMAGE out int status int func IMAGE IMAGE IMAGE pointer to function im_begin_func add_im find function pointer for type of inl func overload func add im inl if not present for this image type report the error if func return im_report_error adq im IE_NOTOVERL status func inl in2 out perform operation if error detected exit function with error value if status lt IE_OK return im_report_error add_im status normal exit im enq func add_im return IE_OK The function overload_func returns a pointer to the function that performs the add operation for the type
145. e use the C syntax to denote AND with amp Replacing amp with l denoting the bitwise OR in algorithm 1 gives the set union Note that not only the image itself is processed but also the border This small amount of extra work makes nested for loops superfluous ALGORITHM 1 Logical Operations Shown is the pseudo code to perform the logical AND of two images represented in bitmaps B and C resulting in bitmap D 11 6 Bitmapped binary images Image 2 1 Library User manual FOR i 0 TO MaxIndex DO D i Bfi amp C i ENDDO Implementation of the Morphological Operations Rather than operating on one central pixel in combination with neighboring pixels the bitmap organization of binary images allows us to operate on 32 central and 32 neighbor pixels in parallel This is simple in case we operate on the central pixel and the pixel above or below because if A i is the word containing 32 central pixels AG LINEOFFSET and A it LINEOFFSET contain the 32 pixels just above respectively just below the central pixels However if we operate on a central pixel with the neighbors to the left or to the right it gets more complicated Again assume that A i contains the 32 central pixels then the pixels just to the left of these central pixels are not in one word in the bitmap The left neighbor of the leftmost pixel in A i is in the word A i 1 In order to represent all the 32 left neighboring pixels we introduce
146. e E a an ae nias 9 16 Only OUDOT aa EE EEE R EE E cae Ted EE EAS 9 16 Type of input sizes of output ssesssesssesessseesseessessseesseeesseeesseesseesseesseeesseee 9 17 Special SIZES cesia Es E edad a E a a a 9 17 BC AN POS O a a a e e E N Gaccac taeda 9 18 Miultiple calls 10 pre OPs n iii je Menta BA ae ASS 9 18 The post op AMNCWON niesi ti seas ine R aE aea AET EEAS tered 9 19 ROLard post Opisercnne aa a e aa a as 9 19 Asimpleexa mp Ea sm Re Ben E RE ie POP POE DOE A AA aA ia aeS 9 19 Explanation of the function code seesseeseeseseseseesseseresresstseresresserseesereseeseeseessee 9 20 Error handling and Ve porting 9 su sascha sencossossaesade wag atagetecs tate qsdes ua geeesesoe Ma eee 9 21 Location Of the errorae eean R a a ee eee Aan ee eee 9 21 Return values of functions 73 4 aetna leccste eae as nee alee eset 9 22 Error atin gio cele dae eo acai Saad ee ge et Been ee Rees a 9 22 Checking TOUINES os ccsze corsets desea a aaea e aa aa a a Ta tinea 9 23 Check functions nasienie e ea a a satin tee peated 9 24 Check image mte rity i an pa E a ai A R EAEE OAE aaie 9 25 IRSE EDONOLA 9 11 Reet ner Mn Eat att tesco nt ROnCRE PRE E E 9 26 F ncton overloadih assie e e ia eats ace tage ao el ned eee aaa 9 27 IR aine E VAN E N OE N tect de ies ated ames ts Segoe S Mae Ae 9 27 G nerie PUNCH ON layette A endnote eet 9 28 Parameter checking and image adjustMent ceecceceseeeceeeceeseeeeeeneeeenaes 9 28 Processing GG
147. e Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual Expression evaluation on images eval A number of times in this chapter the command eval has been used Eval is an expression evaluator for images This means that an operation on an image can be given in the form of an expression that is then executed on each pixel in the image For instance eval A 100 sets all pixels of image A to the value 100 Let us try an operation on a real image readfile house A eval B A 2 80 The value of each pixel in B is determined by multiplying the corresponding pixel in A by 2 and subtracting 80 Eval can handle both GREY_2D GREY_3D FLOAT_3D and FLOAT_2D images The type of image the result is stored in depends on the expression the command given above all result in GREY_2D images but eval C 123 45 results in image C becoming a FLOAT_2D image If the result of the expression is an integer then the result will be a GREY_2D image or GREY_3D if the images used are 3D images likewise if the result is a floating point number then the image will be a FLOAT_2D image or FLOAT_3D The expression can be very complex because eval can handle all C language operators as they are listed here with their priority operator associativity priority oL left HIGHEST right se Jo left amp amp left gt gt lt lt left gt lt lt gt left l left amp left A left l left amp amp
148. e RunEditor by clicking the close box of its window To run your macro file use the macro command which has the following syntax macro i v lt macrofile gt The macro command executes the lines in the macro file as if they were typed in one by one The contents of macro files are restricted to lines with the syntax of the Direct Command Mode A macro file is not a function it can have no arguments As an example suppose it is frequently necessary to clear four standard images named A B C D Instead of typing the command clear_im four times a macro file can be created Edit a file using File New C Program to contain lear_im A lear_im B lear_im C lear_im D COO One Save the file using File Save as cleanup mac recall that the suffix mac in the filename is meant to indicate a macro file Now we have created a macro that clears all images First we read an image file into each image window readf trui a copy_im a b copy_im b c copy_im c d Then we clear them using our newly created macro macro cleanup mac The macro command has two optional arguments i Each line is executed with user interaction y Each line is printed to the Worksheet window verbose When you use the v option please note that the individual commands from the macro appear in the Worksheet but that the History window only contains the macro command 4 6 The C Interpreter SCIL_Image 1 4
149. e already no new memory will be allocated and the input image descriptor and the Programming with Image 9 15 Image 2 1 Library User manual output image descriptor remain the same When an operation is performed in place input image is the same as the output image the pixels will be replaced one by one as soon as each of them is processed No reference to an original pixel value can be made once it is processed For most operations this is not a problem Neighborhood operations however cannot be performed in place To allow the same image to be specified as both input and output for a neighborhood operation pre_op can be used with ADJUST_NIP ADJUST Not In Place If ADJUST_NTP is used the output image will be split up if the input image and output image are the same even if no change of type or size is required When this option is used the output image will always be a different piece of memory and the neighborhood operation can be performed with the same image as both input and output if necessary Output equal to input The most common call to pre_op is for adjusting the output image to match the type and sizes of the input image For instance in order to copy an image to another image the output must be of the same type and sizes The call to pre_op to achieve this would be pre_op in out ADJUST G_2D_SPEC G_3D_SPEC OUT_AS_IN In this call first the image in is checked if it is either
150. e created to tell the infrastructure which function to call for which basic service Overload tables in the section Programming with Image describes the overloading mechanism and the format of the overload table The source file overload c is generated from the overload tables as described in that section When the file overload c is compiled thet ype_ident and type_spec of all valid image types must be known The header file image h is included in overload c to provide 12 14 New image types Image 2 1 Library User manual these definitions for the standard image types The preprocessor directives include and define present in an overload file will be copied to the file overload c By including the header file with these definitions in the overload file the new image types are made known to the overload mechanism For the byte 2d image type construct a file byte_2d ovl with this contents note the include directive include byte_2d h BYTE 2D OVERLOAD TABLE include byte_2d h ABLE byte2d BYTE_2D BYTE Basic functions for the byte_2d __2D_SPEC image type te_2d_create_image te_2d_destroy_image create_image by destroy_image by copy_part_image by copy_masked_part by te_2d_copy_part_image te_2d_copy_masked_part 2d_conv_to_common 2d_conv_from_common 2d_pix_value_str 2d_get_image_window_info
151. e type specific type descriptor typedef struct byte_2d_t int type mandatory field BYTE data mandatory field int lenx int leny BYTE_2D_IMAGE Macro s to allow for access define Byte2dImageType bp bp gt type define Byte2dImageData bp bp gt data define Byte2dImageWidth bp bp gt lenx define Byte2dImageHeight bp bp gt leny endif BYTE_2D_H The prototypes of all the function for the image type are kept in a separate include file pyte_2dp h Its contents is listed in the Low Levels paragraph on page 12 15 Creation and destruction The first two functions needed for a new image type are the create and destroy functions The infrastructure uses these functions to create this type of image and to dynamically adjust the type of an existing image Every image processing operation specifies the type and sizes of the output image These functions are called by the infrastructure to adjust the image to the operation s demands as required The task of the create image function is to allocate a type descriptor the type description structure the space needed for the image and to fill in the structure A pointer to the newly created type descriptor must be returned on success When either of the two allocations fail the already allocated space if any must be freed and a NULL pointer must be returned Function overloading requires the func
152. e what you want The reason is thatexit is a standard C function known to SCIL_Image which interprets it as a command to quit You will get the functionality of exit by using return inamain or by using S_error 0 Even when a program is syntactically correct it may still contain a programming error only detectable on execution Such program errors may originate from illegal memory references illegal instructions or floating point exceptions to name a few If one of these errors occurs the SCIL_Image interpreter will attempt to stop execution print the corresponding error message and return to the Direct Command Mode For example if an attempt is made to redirect a NULL valued pointer execution of the following for loop will be stopped enabling the user to correct the mistake int ptr for i 0 i lt 100 i printf d ptr i for i 0 i lt 100 i printf od ptr gt indirection of a NULL valued pointer The other program errors generate similar error messages However if your error is sufficiently severe it may crash SCIL_Image This will delete all changes you have not saved and destroy all unsaved images You may even have to reboot If you are working in programs you do not trust it is always wise to often Save files and relevant images Also use logon see Program Development Commands page 4 9 to keep a record of the commands you gave Features of SCIL_Image s
153. e your new interpreted function with standard SCIL_Image The function will appear in menus and is ready for use The Command Description File contains the layout of the menu system and on each available command the information needed by SCIL_Image to execute that command properly Modifying the Command Description File is appropriate is you wish to hide some or the 1 4 Outlines of SCIL_Image SCIL_Image 1 4 User Manual majority of the Image functions from your users sight Or may you wish to re order the image processing functions differently over the menus more suited to your specific needs This can all be done by editing the Command Description File Making your own library When a algorithm has been fully developed and tested the set of commands which constitute the algorithm used can be converted to C routines with little effort In general this only requires the addition of a subroutine header and tail A new routine can be added to SCIL_Image by compiling it with a standard C compiler and inserting the routine in your own library At the same time the name of the function and the parameters and the admissible parameter values must specified in the Command Description File to make it acquainted to the system When this has been done correctly the function will be available to the user in the same way as any of the standard image processing functions Building applications For a specific application it is ofte
154. e your own manual pages For this you need an editor that can read this format After you are done Advanced SCIL_Image 5 17 SCIL_Image 1 4 User Manual composing your Help page according to the instructions your editor window should look similar to this w Fie Edit View Insert Format Tools Table Window Help dg gt tl aae Ee Buga m zal K Heading 1 7 FS add on amp SCIL NAME NAME Normal add_one add 1 to an image Normal q SCIL SYNOPSI SYNOPSIS ScilCourier include user def h ScilCourier include image h Y Normal 1 ScilSyntax int add_ one IMAGEimage h in IMAGEimage h out 7 Normal q SCIL DESCRIP DESCRIPTIONY Normal Add 1 to the image in and store the result int the tmage out J Normal 1 SCIL RETURN RETURN VALUESY Normal OK 1 ifsuccessful Normal NOT_OE 0 ifnot successful Normal 1 SCIL AUTHOR AUTHOR Normal written by M Y Self 199702049 Normal 1 SCIL ENDREF fan Footnotes Close Footnote Text add_onefl Footnote Text S add_onefl Footnote Text Eadd_one MY FUNCTIONS Footnote Text USER_FUNCTIONS 000024 ba EKIN nfs Page 1 Sec 1 v1 At Ln Col 1 REC MRK EXT GVA WPH Save it and start the Help Workshop from the Microsoft Visual C environment In the Help Workshop open the file scilimag hpj from the help directory To compile the help file Select File Compile
155. ec of image out and is therefore always correct The pre_op calls will therefore accept every type of image Type of input sizes of output In the above examples using pre_op in ADJUST mode the output image inherits the sizes of the image that is specified as the first argument This is however not always desired in which case special measures have to be taken For instance if the output image must be of the same type as the input image but keep its own sizes the next two calls can be issued to achieve this behavior pre_op in in COMPARE G_2D_SPEC OUT_AS_IN pre_op out out ADJUST ImageTypeSpec out ImageTypelIdent in The first call checks if the input image in is of the correct type in this case GREY_2D The second call to pre_op converts the output image out to the type of the input image in by specifying ImageTypeldent in as the type to which the image out must be converted Because the image out is specified as the first parameter the sizes of out will be used This results in the image out being converted to the type of image in with no change in size Special sizes When the required sizes of the output image differ from those of both the input and output images a special method is available to change the sizes The function set__cross_dim can be used to tell pre_op that the next ADJUST action must use the supplied special sizes The following call t
156. ecks only for those pixels in a structuring element for which Ill gt 1 This choice is no essential restriction due to the chosen border and can be easily lifted at the expense of extra border checks Although structuring elements are sets and could have been represented with bitmaps we have chosen another representation which can be more efficiently dealt with in the algorithms described in the next subsection Figure 11 3 depicts the data structure used to represent arbitrary structuring elements The indicator array F is used to represent a hit or miss mask S T in one structure By definition a pixel with index i is element of S if F i gt 0 and element of T if F i lt 0 The it element of a hit or miss mask is described with two coordinates X i Y 1 and its indicator F 1 STRUCTURING ELEMENT DESCRIPTOR a b Figure 11 3 Hit or miss Mask Representation In a an example of a simple hit or mask S T is shown with pixels in S denoted as 1 and those in T as 1 In b the corresponding mask representation is shown Implementation of the Pixelwise Logical Operations The pixelwise logical operations of one ore two binary input images are the Boolean equivalents of the union intersection and complementation of sets Using bitmapped binary images these operations can be performed very fast as 32 pixels are operated upon in parallel Algorithm 1 shows the pseudo code to perform the logical AND of two binary images w
157. ect that wishes to monitor the changes of that image must subscribe to it using the spb_subscribe function The current messages that are published by image objects are 8 6 Publish and Subscribe Image 2 1 Library User manual SPB_DESTROY is published when an image is about to be destroyed This publish is done right after the SPB_ITEM_DELETE message of the super_im It describes the same event of the same image It also is published prior to the actual destruction of the image which cannot be halted The data with this message is a NULL pointer because the image is identified already by the publishing object first parameter of the subscriber s function SPB_CHANGED is published when the image data changed The data with this message is a pointer to a SObAREASD structure This structure describes the region that changed the offset and the size A NULL pointer is supplied when the entire image changed This happens with most image processing operations SPB_NEWSTATE is published when the Image Flags changed The data is a pointer to a SObNSTATE structure that contains both the old and the new value of the Image Flags SPB_NEWCLUT is published when the color lookup table CLUT of the image changed or was replaced by another The data is a pointer CLUT to the CLUT SPB_NEWTYPE is published when the type of the image changed The data is always a NULL pointer The new type can be retrieved from the im
158. ectangular ROI image Especially histogram operations and statistical analysis on these pixel values may unknowingly be influenced by the 0 pixels Operation Counter The op_cnt field of the IMAGE structure is a counter that is incremented each time the image is used as an output image for an image processing operation At several places in the image infrastructure this operation counter is used to test the validity of image related data For instance the image data of a ROI is copied only to the ROI image if the op_cnt of the parent image indicates that the data of the parent image has changed The post_op function is responsible for incrementing the counters so image processing functions that do not use pre_op post_op on their output images can result in inconsistent ROI data Programming with Image 9 9 Image 2 1 Library User manual Image Info The image_info pointer in the IMAGE structure can be used to store arbitrary additional image related data with an image By storing the data with the image it is not needed to keep track of which data belongs to which image A linked list of IMAGE_INFO structures is attached to the pointer Each structure in the list points to user application specific data The data stored with each element of the list is neither interpreted nor processed by Image in any way The user application remains responsible for the contents of the data as well as the allocation of the memory if
159. ed C functions Chapter 5 e Putting commands together More than one direct command may be entered on the same input line if separated by semicolons For example readf trui printf There she is again n The first semicolon is thus a separator between direct commands the second semicolon has to be given since the second command is a C function call and is not syntactically correct without it 4 4 The C Interpreter SCIL_Image 1 4 User Manual Using the Worksheet and the History window you can work very flexible in the Direct Command Mode as our Sample Session Three page3 21 showed Both the Worksheet and the History window show previously entered commands The difference is that the History window cannot be edited it retains an exact account of what you have done But the Worksheet can be edited which is convenient if you want to type a similar command to one you did before It uses the standard Windows editor available under the Edit menu Both windows can be resized moved and scrolled We repeat the main possibilities of these windows e Ifyou want to execute more than one command in the Worksheet or in the History window you select them by using the click and drag method until all desired commands are highlighted then execute them hit the Enter key on the numeric keypad or Shift Enter on the keyboard e Ifyou want to re execute a command you can do this simply by moving the cursor to
160. ed by applying the OR operation on the type_specs of the classes see Table 9 1 present image types This is useful when the same function can be applied to more than one image type When the type of the image first is not important the define WHATEVER can be used all bits set The last parameter second_ident specifies the intended type_ident of the image second This image then will be converted to that type if ADJUST ADJUST_NIP is specified and must be of that type if COMPARE is specified This parameter may be specified as OUT_AS_IN the meaning of which depends on the mode as explained below 9 14 Programming with Image Image 2 1 Library User manual COMPARE mode When mode is specified as COMPARE pre_op determines whether two images are of the same class and or dimensions It can also be used to check if one image is of a specific type For instance when adding two images it is essential that they are of the same type and sizes The call to pre_op to check this would be pre_op inl in2 COMPARE G_2D_SPEC G_3D_SPEC OUT_AS_IN inl and in2 are the two input images The valid classes are both G_2D_SPEC and G_3D_SPEC because the routine add_im can handle both 2d and 3d images The last parameter is OUT_AS_IN because the second image in2 has to be of the same class as the input image in1 So in this case the allowed types of images are GREY_2D LABEL_2D same clas
161. een and Blue component each component channel occupies 1 byte in memory C type unsigned char The three channels together with a fourth extra and empty byte are stored in one C structure CIE XYZ each pixel consist of three floating point values X Y and Z that are stored in one C structure Each channel occupies 4 bytes of memory C type float CIE Lab each pixel consist of three floating point values Lightness a and b that are stored in one C structure Each channel occupies 4 bytes of memory C type float HSI each pixel consist of three floating point values Hue Saturation and Intensity that are stored in one C structure Each channel occupies 4 bytes of memory C type float CMYK each pixel consist of four floating point values Cyan Magenta Yellow and K sometime also named Blackness that are stored in one C structure Each channel occupies 4 bytes of memory C type float The Image 2 1 library in SCIL_Image 6 17 SCIL_Image 1 4 User Manual One of the reasons for the existence of different color models we implemented only a few of the many that are defined is the suitability of each model for a specific task The capability to convert one model into another model is therefore essential Try readfile flamingo A convert_cmodel A B HSI_T The color model of a color image is printed also in the title bar of the display window When comparing image A and B see that image A reads col2d RG
162. efault implementation or off course replace the existing rule This can be done by placing it later in the same file as a pair with the same message or by placing it in a new file which is lower in the list of overload files from which overload c is created The location of this list depends on the platform you are using Image on it is either in the makefile UNIX amp PC or in the file ovl_list Macintosh Programming with Image 9 31 Image 2 1 Library User manual Data conversion convert Occasionally conversion of the image data is required for example if an operation on grey valued image data must be performed in floating point precision A special mechanism is supplied which enables one function to convert any type of image in Image into any other type Even if new image types are implemented no altering of existing source code is necessary Because the convert command makes use of the function overloading mechanism users can overrule the method used to convert the image data Super type of an image line The routine convert uses the special image line type called a common_line to store a line of image data By having a routine for each image type that puts the data of an image line in acommon_line structure and a routine for each image type that reads a line from the common_line structure any type of image data can be converted to any other The entire image is converted by calling the former funct
163. en in Function overloading page 9 27 The rules for composing noise c and noise h are just those of standard C see any C manual but also beware of the slight deviations of SCIL_Image s C see Features of SCIL_Image s C interpretet page 4 16 You might want to study the example in Making a New Compiled Version of SCIL_Image page 3 30 First an important note on compiled functions in SCIL_Image IMPORTANT Special include file for I O When adding compiled functions to SCIL_Image that use printf fprintf putc etc to print text to the Worksheet or getchar scanf etc to read text and characters from the keyboard the include file scil_io h must be included in the source file Perform the following actions 1 Open the Microsoft Developer Studio workspace c scilimage14 prog scilimag dsw if not already open and add the file add_one to the project mylib right click mylib and choose Add files to project and then select the file noise c from the directory you put it in 5 16 Advanced SCIL_Image SCIL_Image 1 4 User Manual 2 Open de file c scilimage14 prog comfiles mak set the Open as to text in the dialog box and at the end of the COMFILES list add c scilimage14 prog My_src noise cdf and save the file fill in the appropriate path name if you saved the file noise cdf somewhere else 3 Open de file c scilimage14 prog overl
164. erely defines the type of the text and should be considered an aid for the user interface to enable it to present different kinds of text in different ways The default behavior of image_output is to dump all text to stdout except for the IMO_ERROR stream which is dumped to stderr User interfaces can intercept the text by supplying a function pointer that conforms to the following prototype void WINAPI funcname int stream char buffer The image_output function converts its variable argument list to a single text buffer that is given to this function To supply the function pointer use the function include imtxtout h im_set_output_handler void WINAPI int char The defined types of text are definition can be found in the include file image h We use WINAPI on the MS Windows platform to provide greater flexibility On other platforms WINAPI is an empty define see the include file imtxtout h 9 26 Programming with Image Image 2 1 Library User manual stream intended type of text IMO_OUTPUT default text such as measurement results progress information etc IMO_INSTRUCT instructions for the users e g information on how to operate an interactive routine IMO_WARNING warning messages IMO_ERROR error messages Table 9 3 present output streams Please note that although we define different streams of text output we do not impose any behavi
165. es initimage initializes all the necessary internal tables of the library It is very important that the initialization is done otherwise most image processing functions will not operate at all Finally the image processing itself in this sample we choose to leave out checking the return values of each function except for the initial reading of an image In a real application additional statements are needed to check that each operation performed its task correctly and if not take appropriate action de lt stdio h gt de lt string h gt de sSupport h de spublish h de image h de im_error h de generic h de im_infra h de imonly h de imtxtout h de im_aio h incl incl incl incl incl incl incl incl incl incl incl Go iGo G Kee Ge GAG ie Gs static char errbuf 2048 local buffer for printing etc IM_ERROR_STATUS err_stack void super_image NULL int main int argc char argv IMAGE im iml im2 LIST 1 o OBJECT obj get the global objects to subscribe to super_image get_super_im fe PW ey err_stack get_im_error_stack 2 set our own text output handler im_set_output_handler my_output_func 3 subscribe to the error stack spb_subscribe err_stack NULL im_handle_error OL 4 subscribe to image class super_im 13 4 Creating an Image Appl
166. es supplies in depth information about this feature The topic is discussed by means of an example on how to implement byte images 8bit in Image Measurement of objects AIO AIO is a framework for the Analysis of Images and Objects It is designed to perform measurements of microscopical objects gathering results in linked lists These results can than be stored and used in databases e g for statistical research Objects are manipulated either based on the measurements or by pointing in an image when using a GUI The object data vector and the object image can be stored in a file and in an image silo respectively for later retrieval The combination of AIO and Image provides a flexible environment for microscopical image analysis for both experimental use and application development In Analysis of Images and Objects AIO the capabilities of the AIO package under Image are described Binary images In the implementation of the binary image types the data structures used for image manipulation received careful attention resulting in optimal performance for binary operations By packing the binary data in a word we are able to process 32 pixels in parallel An important set of mathematical morphology functions including erosions dilations and skeletons are implemented for arbitrary shaped structuring elements The arbitrary shapes make the mathematical morphology functions useful for a broad class of image processing
167. es the implementation of this function in the SCIL_Image package is supplied as an example Please note that the point_im and MousePress functions are not a part of Image library either Their functionality is discussed further down The point_object function The function is fairly straightforward in a continuous loop the location of mouse clicks are received from the point_im function 1 until a key is pressed c unequal to zero It is then checked if the mouse click is indeed a mouse down click in the correct image 2 The pixel value from that location in the image is retrieved 3 and find_object translates this into the handle to the corresponding object in the object list 4 This handle is returned to the caller 5 which can use it to measure the object using other AIO functions include image h include im_error h include im_aio h LIST point_object IMAGE im LIST link IM_EVENT but int Cpe Ki Y IMAGE ip PIXEL pix LIST lp AABEL_2D_IMAGE tmp im_begin_func point_object while 1 if c point_im amp ip amp x amp y amp but 1 im_end_func point_object return NULL if im ip MousePress but 2 continue tmp LABEL 2D IMAGE ImageIn im pix Label2dImageData tmp 3 long y Label2dImageWidth tmp x if lp find_object link x y pix 4
168. execution of the image processing functions of the Image library When coding an application several topics may need additional attention e What to do when an error occurs in the Image library and how to retrieve information about the error situation See Error handling on page 13 2 e How to redirect text output from the library to the desired location See Text output handling on page 13 3 Initialization Before any image processing function can be called the library must be initialized The function that performs all the necessary actions is init image Its prototype void initimage void Although for a complete application additional steps are necessary to get automatic notification of all kinds of events such as errors creation of images changes to images etc the initimage function is sufficient to get the Image library running correctly Error handling In the event of an error occurring in the application due to whatever reason the program must detect the situation and handle it appropriately Either by trying to correct to problem or quitting the application For correct assessment of the situation sufficient information is required As discussed in Error handling and reporting on page 9 21 the Image library logs the location and nature of an error occurring in the library in detail After a processing 13 2 Creating an Image Application Image 2 1 Library User manual function h
169. fset The function operates on images of type GREY_2D size should be a value between 3 and 12 and offset may be any positive value The function which check its input is as follows int my_filter IMAGE in IMAGE out int size int offset im begin func my_filter if range_ok size 3 12 Filter size positive_ok offset Offset return im report_error my_filter IE_ NOT_ OK error in parameters if pre_op in out ADJUST G_2D_ SPEC OUT_AS_IN return im_report_error my_filter IE NOT_OK NULL rest of function post_op out im_end_func my_filter return IE_OK It is a sound practice to first check the validity of all the non IMAGE parameters and when these are all O K then call pre_op This prevents the need to undo the pre_op action if one of the other tests fails Please note that the error mechanism checks the integrity of all images when an error has occurred In the example the first check is performed with range_ok which checks whether the size parameter is between 3 and 12 If it is not in the specified range e g when size 21 the following text is logged in the error stack Filter size 21 out of range 3 12 The default error message in range_ok is prefixed with the supplied text string Filter size If the check range_ok succeeds the check positive_ok is issued If this check fails this text wil
170. function ARGS char specifies a variable argument list In all other cases each argument must be described as follows lt tab gt The argument list for a function continues until the first line which does not start with a space or tab A lt tab gt is often used to improve readability Any line starting with white space is assumed to contain an argument description Therefore you should not have any empty lines at the end of your CDF file type The type field specifies the argument type The admissible types are in two categories C variables and the SCIL_Image special types The C variables are int uint short ushort long ulong float double char uchar uint stands for unsigned int but since only a single word can be used in the CDF it has to be abbreviated ushort ulong and uchar are similar The SCIL special types are text pass argument unaltered string text surrounded by optional double quotes odd odd integer value only even even integer value only confirm Yes No 1 0 Advanced SCIL_Image 5 9 SCIL_Image 1 4 User Manual switch On Off 1 0 choice one choice out of a list of choices toggle multiple choices out of a list of choices filename ask for name of a file filesave ask for file name by popping up savebox A more detailed description of the SCIL special types is given in SCIL_Image Special Types page 5 10 vis This field is reserved preferably it should contain a minus si
171. ge Environment SCIL_Image resides on your computers in directories of your choice You have to tell SCIL_Image what these directories are When you have just installed the system everything is where it is expected and you can immediately start with the sample sessions in Chapter 3 Later on you may want to move everything somewhere else You then have to edit the file scilinit in the scilimage14 directory You can do this from SCIL_Image using the File Open C Program menu entry which contains a standard editor ee RunEdit C scilimage1 455 cilinit printf Welcome to SCIL_Image 1 4 Windows n SETENU PICT c scilimage14 images SETENU HMACRO c scilimage14 prog include c scilimage14 demo 5c scilimage14 demo macro SETENU SCIL c scilimage14 SETENU SCIL_HELP scilimage14 help scilimag hlp There are four variables that tell SCIL_Image where its directories are on your computer You have to use these variables to tell SCIL_Image where to find your data and programs These variables are called environment variables PICT The folders where images are kept At delivery this is set relative to the directory in which you chose to install SCIL_Image so in our example c scilimage14 images MACRO The folders which contain programs macros and include files for At delivery set to c scilimage14 prog include c scilimage14 demo c scilimage 14 demo macro
172. ge processing Knowledge of the C language is not required for this chapter SCIL_Image 1 4 User Manual Introduction In chapter 1 it is explained that the image processing functions all reside in the Image library which can be used independently from the SCIL_Image environment In order to use Image effectively for image processing it is necessary to know something about its infrastructure This chapter describes the Image infrastructure and the SCIL_Image specific user interface on Image Any individual image processing commands that are mentioned are documented in the reference manual s help file One of the features of Image is the support of various image types The image types implemented in this release are image type name in Image _ intitle bar grey valued 2 d GREY_2D g2d binary bitmapped 2 d BINARY_2D b2d floating point 2 d FLOAT_2D f2d complex 2 d COMPLEX _2D c2d color 2 d COLOR_2D col2d labeled 2 d LABEL_2D 12d grey valued 3 d GREY_3D g3d binary bitmapped 3 4 BINARY_3D b3d floating point 3 d FLOAT_3D f3d complex 3 d COMPLEX _3D c3d color 3 d COLOR_3D col3d labeled 3 d LABEL_3D _ 3d Table 6 1 image types available in Image In Image the user does not need to know the type or dimensions of an image an operation will result in The package automatically sets the size and type of the output image based on the operation and the type and size of the input image This adjustment also
173. ges SPB_ITEM_ADD is published when a new clut object is created The data is pointer CLUT to the new clut SPB_ITEM_DELETE is published when a new clut object is created The data is pointer CLUT to the clut When this message is published nothing can be done to stop the destruction of the clut The only reason this message is published beforehand is to prevent invalid memory access by subscribers that still reference the clut because they do not yet know that the clut no longer exists SPB_CHANGED is published when the contents of a clut is changed The data is a pointer CLUT to the clut All functions that somehow change the contents of a clut must publish this message Top level Histogram publishes Like the images the histogram objects publish events to allow for a flexible way of handling histograms Similar to images a hook is permanently present to provide a way to detect the creation of new histograms This top level histogram is called super_histo and is globally available as a void pointer or via the function get__super_histo void s_histo s_histo get_super_histo spb_subscribe s_histo NULL funcname NULL The messages the top level currently publishes are SPB_ITEM_ADD is published when a new histogram object is created The data is pointer HISTOGRAM to the new histogram SPB_ITEM DELETE is published when a histogram object is about to be destroyed The data is a pointer to the c
174. gn default This field holds the default value of the argument The default may be a string or a SCIL variable for a SCIL variable the current value is substituted When a command is issued with too few arguments defaults are used for the missing parameters The default values are also presented in the dialog box min max The interpretation of the min and max fields depends on the argument type For numerical types the value of a given function argument is checked against the legal value range specified in the min and max fields If a value is outside the range SCIL_Image will complain with the message out of range The user will then be prompted to enter a value within the legal range An isolated minus sign may be entered in the min or max field in that case SCIL_Image does not check your entries Many of the SCIL special types use the min and max fields for special purposes This is described in in SCIL_Image Special Types page 5 10 prompt The prompt is used to label fields in the dialogue box it is the text that will specify what needs to be entered It also appears on the screen if a question mark has been entered or if an invalid value has been entered in that case it includes the display of the default within brackets If the user enters too few parameters followed by a question mark the prompting will appear for each of the missing parameters SCIL_Image Special Types We saw in the previous section that
175. he decompositions of 8D The well known and often used linear decomposition gives the total number of pixels in the structuring elements is 40 efi pinikit t tE Thus instead of eroding dilating with a large 17x17 structuring element we can erode dilate 8 times using a 3x3 structuring element The linear decomposition using the extreme set E D gives total number of pixels considered is 33 vki gp p pi i tt The logarithmic decomposition finally gives total number of pixels considered is 17 Bitmapped binary images 11 3 Image 2 1 Library User manual It can be proven that logarithmic decomposition can be used for all convex structuring elements For the 4 connected and 8 connected neighborhoods logarithmic decomposition is automatically taken care of in Image Algorithmic Implementation Data Representation In most modern general purpose computers the Von Neuman bottle neck still exists The efficiency of low level image processing operations is bound by the speed at which the image data can be transported between the memory and the CPU Traditionally in image processing binary images are stored as one bitplane in a NxM pixel mapped grey value image In a pixel mapped image the grey value of each pixel is stored in one computer addressable memory word A binary image needs only one bit per pixel to indicate whether that pixel is part of the object or whether it is a background pixel To access the va
176. he ROI itself CANNOT be adjusted if the ROI itself is specified as an output image Consider the following readfile trui A roi_define roil A 30 50 0 128 1921 copy_im roil B 6 10 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual A region of interest has been defined that starts at x 30 y 50 with width 128 and height 192 The command copy roil B copied the part of image A defined as roil to image B Image B therefore is adjusted to match the dimensions of the ROI rather than those of image A Let us see what happens when roil is used as the output image for a command Try the following threshold B roil As a result of this command the image type of the ROI must be changed to binary However the ROI is simply a geometrical part of image A and therefore image A must be changed to a binary image to store the result of the operation in the ROI Only the ROI contains the result of the operation the rest of the image is set to zero A ROI need not be rectangular but may be arbitrary shaped see the on line manual of roi_define Image types Grey valued images GREY_2D amp GREY_3D Data representation of grey valued images Grey valued images consist of pixels each of which contain an integer value 1 2 These pixels are represented in SCIL_Image by a in the C language defined datatype PIXEL This type is defined to bea short int which means that on most computers 1
177. he binary image B Because the threshold requires a grey valued input image the system pops up an alert box with the text WARNING N SCIL_Image Error Function threshold Message Function is not implemented for image type BINAR Y_2D E Stop More info Figure 6 1 Alert box on incorrectimage operation The first line SCIL_Image Error indicates that this error message is generated by SCIL_Image The next line Function threshold shows the name of the function that caused the error and the line starting with Message describes the error At the bottom of the box are three buttons Dismiss removes the alert box and continues with the next statement if any in an interpreted program Stop also removes the alert box and stops running the interpreted program if one was running In this example only on command was given so the difference between Dismiss and Stop in not noticeable The More info button shows a new alert box that displays a stack trace of the location of the error That info is generally most of interest for those that write new image processing functions The Image 2 1 library in SCIL_Image 6 3 SCIL_Image 1 4 User Manual Image display and window management An important part of interactive image processing is the display of the image Several commands are available for this purpose as can be seen in the Display menu Please note when reading this section the distinction be
178. he trick T int byte_2d_conv_from_common IMAGE im int n COMMON_LINE com_line BYTE_2D_IMAGE Kois double d_ptr long ZLPE BYTE pix int type register long npix register int chan_offs im_begin_func byte_2d_conv_from_common g_im BYTE_2D_IMAGE ImageOut im New image types 12 13 Image 2 1 Library User manual type com_line gt type d_ptr com_line gt data l_ptr com_line gt data chan_offs com_line gt nr_channel npix Byte2dImageWidth g_im pix Byte2dImageData g_im pix com_line gt y npix F r if com_line gt z if type COM_LONG while npix gt 0 pix t l_ptr l_ptr chan_offs else if type COM_DOUBLI while npix gt 0 pixt d_ptr d_ptr chan_offs Gl a else if type COM_LONG while npix gt 0 if pix lt sI pry pix l_ptr pixt l_ptr chan_offs else if type COM_DOUBL while npix gt 0 if pix lt d_ptr pix d_ptr pixt d_ptr chan_offs Gl if com_line gt t gt 0 im_report_error byte_2d_conv_from_common IE_NOT_OK return 0 0 stop can t handle this im_end_func byte_2d_conv_from_common return 1 1 can handle more data The overload table When all the basic functions have been created an overload file must b
179. hoice in Object Object 1 Background 0 Make Contour of In this example note that because there are only two valid choices with values 1 and 0 a switch could have been used rather than a choice with the same flexibility in defining the terms displayed in the dialog box In this case it is primarily a matter of preference whether a switch or a choice is used as the argument type toggle resembles the choice type in many ways The difference between the two is that 5 12 with a toggle you may choose more than one selection from the given list whereas the choice type allows only one selection The syntax for the toggle type is exactly the same as the choice type SCIL_Image performs a logical OR on the choices and calls the function As with choice you can attach aliases for easier readability An example Advanced SCIL_Image SCIL_Image 1 4 User Manual im_aio cdf UNC measure void ENU Single_Objects ARGS image in A 7 Grey image image in B Binary image int con 0 0 262144 Garbage level confirm con No 7 Interaction toggle con AREA PERI ARFA 1L PERI 2L CR 4L BE toggle con AREA PERI ARFA 1L PERI 2L CR 4L BEND 32768L XMIN 8L XMAX 16L YMIN 32L YMAX 64L WIDTH 128L HEIGHT 256L GRAVX 512L GRAVY 1 0241 ANGLE 2048L Shape toggle con GREYVAL GREYVAL 1L TRANSMIS 2L OD 4L Density confirm con Yes Print results filesave con
180. ht mouse button Macintosh hold down the Command key while pressing the mouse button as described below Another is to use the command next_plane C Up to display the slice below and next_plane C Down to show the one above Finally the display slice can be set explicitly set_display_slice C 8 Note in the title bar that slice number 8 is now displayed The right mouse button The right mouse button is currently only used for the manipulation of 3D image display As stated above 3D images are displayed one slice at the time If the right mouse button is pressed when the pointer is in the upper third of the display window then the next slice of the image will be displayed For instance if slice 0 was displayed slice 1 will be displayed after pressing the right mouse button If when the pointer is in the middle third of the window the right mouse is pressed then the same slice will remain displayed Finally if while the pointer is in the bottom third of the display window the right mouse button is pressed the previous slice will be displayed For instance if slice 5 was displayed then slice 4 will be displayed after pressing the right mouse button This behavior can be disabled with auto_plane No and enabled with auto_plane Yes 6 6 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual Changing a window s size In order to investigate particular image details it may be useful to enlarge the w
181. ication Image 2 1 Library User manual spb_subscribe super_image NULL handle_super_im OL 5 initialize the image library initimage 6 im readfile cermet NULL 0 0 if lim LE ETE image_output IMO_OUTPUT Can not find image exiting n return 0 iml create_image im1l BINARY_2D 256 256 1 threshold im im1 128 invert_im iml iml eval cermet 255 cermet 0 im2 create_image im2 GREY_2D 256 256 1 1 list_label iml im2 8 20 ForAllElements o l obj OBJECT Info o image_output IMO_OUTPUT Label 41d obj gt label object_shape_meas im2 o AREA PERI CR image_output IMO_OUTPUT Area 6ld Peri 8g Cr area o peri o cr o return IE_OK JE This function retreives the information from the error stack and puts it in a textbuffer wa void my_dump_err_stack IM_ERROR_STATUS imerr char errbuf Ing i gt char ptr fptr char ep sprintf errbuf Image error stack trace n n n ep errbuf strlen errbuf 8g n for i 1 i lt imerr gt edepth i for each level pointer to message ptr imerr gt mstack imerr gt moffs i 1 pointer to funcname fptr imerr gt fstack names imerr gt fstack noffs i 1 print funcname and error status sprintf ep Ss d fptr imerr gt sstack i 1
182. igure 9 2 Note that this is exactly the opposite of what post_op does Therefore this function should not be called before a call to post_op unless the intention is to restore the state of affairs prior to the call to pre_op and exit the function immediately The function heading is Programming with Image 9 25 Image 2 1 Library User manual void check_image_integrity int print The argument of check_image_integrity tells the function to perform its work silently 0 or print what it is doing 1 This function is called by the Image infrastructure just before it publishes the occurrence of an error Textual output Because the Image Library is designed to be independent of a user interface text output cannot be written to a standard output stream because that may not exist Also the User interface might like to be in control of the representation of textual output For these reasons Image functions should not use the standard printf and fprintf functions for text output to a terminal the fprintf may still be used for specific file output of course The Image library contains a special function for text output the image_output function The prototype int image_output int stream const char format The syntax of the function and the parameter list is like the ANSI C fprintf function except that the st ream parameter does NOT indicate a C F ILE streams It m
183. im im_end_func byte_2d_part_image_display return IE_OKR if ImageDispClut im is_clut ImageDispClut im if no clut ImageDispClut im standard_cluts GREY_LUT_T l1_byte_2d_disp_8bit im ImageVport im sx sy width height im_begin_func byte_2d_part_image_display return OK bit display_routine only scaling no dithering l1_byte_2d_disp_8bit im vport sx sy width height E xim PORT vport sx sy width height short dispw disph unsigned char swap int part_sx part_sy part_width part_height short xsize ysize short xtab ytab unsigned long dtab 256 BYTE_2D_IMAGE ip register BYTE pix register short AXE register BYTE vdi register int Vy register int ane register BYTE tmp im_begin_func l_byte_2d_disp_8bit sizes of the viewport dispw ViewportWidth vport disph ViewportHeight vport get the pointer to the swap area swap unsigned char ViewportSwap vport the sizes of the image ip BYTE_2D_IMAGE ImageOut im xsize Byte2dImageWidth ip ysize Byte2dImageHeight ip copy the display values from the color lookup table for i 0 i lt 256 i dtab i ViewportClut vport gt table i allocate space for the size stretch tables xtab short calloc dispw sizeof short ytab short calloc disph sizeof sho
184. in and the sizes of this rectangle e It must write the data to a swap area of the IMAGE structure This swap area used for handling expose events is allocated by the infrastructure when make_image is called a pointer to it is located in the viewport structure that is attached to the IMAGE structure 12 8 New image types Image 2 1 Library User manual It must be able to handle any window size both the x and y dimensions may differ from the image dimensions When the window is resized the infrastructure will free the swap area and allocate a new one to match the size of the window It must call the function PwWrite to actually display the rectangle written into the swap space The display routines of the standard image types in SCIL_Image handle Intensity scaling which can be set to different modes as specified in the set_display_mode function The scaling mode of an image is stored in the mode_flags field of the VIEWPORT structure Its value is defined in the dmodes h header file Various monitors namely bit monochrome 8 bit color and 24 bit full color monitors Macintosh only 8 bit color is supported Color lookup tables CLUT which can be attached to an image The set_dither_mode function can be used to improve the quality of the display of colors on a 8 bit monitor The dither mode flag is also a member of the mode_flags field of the VIEWPORT structure The function header of the
185. in the pixel in image out that corresponds with the central pixel in the moving window This approach tends to avoid the sub windows with large variations in pixel values e g due to the occurrence of an edge RETURN VALUES IE_OK 1 on success Negative error status on failure see im_error h SEE ALSO edge preserve Figure 3 9 A reference manual page You can browse through this Help facility in the usual way using Contents Search links which are underlined or the gt gt and lt lt keys if enabled 3 14 Getting started SCIL_Image 1 4 User Manual For the next illustration of SCIL_Image we use the fast fourier transform which computes the frequencies in an image Even though this transform uses a fast algorithm it still takes quite a while on a standard size image So we reduce the image first select Image Manipulation change_image_size choose Image B choose New Width 64 choose New Height 64 click OK select Image Manipulation warp_image choose Input Image A choose Output Image B click OK Image A was an integer valued image consisting of 256 256 pixels its header is A g2D 256 256 image B consists of only 64 64 pixels its header is B g2D 64 64 The warp_image command performed the conversion Note that the much reduced image B looks small in its viewport We can display it much bigger by using the zoom an
186. include byte_2d h IMAGE im_a im_b im_c IMAGE roil BOOL_MASK bm create two grey images then from the new type a compare it with the orig HF FUNCTIONS USED create_i xf im_a create_i readfile orka2 im_b create_i convert im_a im_b BYTE_2D convert im_b im_a GREY_2D writefile im_a result1 I GR mag A 56 ima 07 mage B GR create image of new typ images to it To check t save it to disk FUNCTIONS USI ED HF 12 18 create_image im_d read a file from disk convert it to and nd save it to a file again to be able to inal image data mage conv_to_common conv_from_common EY 2D 0 BY 2D CS_F 256 256 1 256 256 1 copy a part of the previously used he result convert it back to grey and copy_part_image conv_to_common New image types Image 2 1 Library User manual im_c create_image C BYTE_2D 256 256 1 copy_part_image im_b im c 64 64 0 128 128 1 0 0 0 convert im_c im_a GREY_2D writefile im_a result2 ICS_F create a Boolean mask from a grey_2d image convert the grey x image to the new type Define a roi in the new type with a Boolean mask and convert that the date from that ROI to a different image type FUNCTIONS USED create_image conv_from_common destroy_image x copy_masked_part afr readfile
187. indow in which an image is displayed Likewise to save display space it can be beneficial to display images in smaller windows If a window is resized using the tools provided with the window system SCIL_Image will scale the view of the image accordingly The size of an image display window can also be modified using SCIL_Image Note that in both cases the dimensions of the underlying image are unchanged Only the display is modified The command set_window_size is used to change the size of the image display window set_window_size A 512 512 To set the window back to the size of the image the same command can be used with the real size of the image shown in the title bar However the command natural_window_size A sets the window to the size of the underlying image in this case 256 by 256 pixels Changing a window s position The window s position on the screen can be manipulated in the same way as the size of the window That is either by dragging the window or by using the SCIL_Image command set_window_pos can be issued For example set_window_pos A 300 0 Display lookup table Often it is useful to modify the colors used for image display without altering the data in the image The color lookup table mechanism enables users to select colors for image display Several standard color lookup tables are created when SCIL_Image is initialized These tables can be attached to an image to set a group of colors to be for dis
188. int_max maximum value in data unsigned int type_ident ident of data source unsigned long type_spec spec of data source COMMON_LINE e type identifies the C type in which the data is stored COM_LONG is the long word format and COM_DOUBLE is the double format e data is a pointer to the allocated memory used to store the line of data see Source function specification on page 9 33 e xis the width of the image e y and z are the position of the line stored in the data e t is the time position of the line not used in current image types e nr_channe1 is the number of values used for one pixel For example grey value images are one channel images and complex images are two channel images e hint_min and hint_max can be used to store the minimum and maximum value of the source image This information can be useful in deciding how to convert the data in the receiving routine e type_ident and type_spec are the type_ident and type_spec of the source image When convert is called the functions for writing to a common_line for the source image and reading from a common_line for the destination image are overloaded Each is called several times until the image is completely converted Each function must have a specific behavior to guarantee the correct conversion of the data Source function specification The header of the source function is specified as follows int source_function I
189. involved in the operation two calls to pre_op have to be made The first one is to s ee if the two input images are of the same type and size The second one is to adjust the type and size of the output image The address of the first pixel of each of the images is acquired through a macro see The IMAGE structure The addresses for the data in the two input images are accessed by the macro Im ageInData and the address of the data in the output image is accessed by the macro 1 mageOutData The reason why the output is accessed by a different macro is described in Dynamic adjustment Pre_op Post_op The number of pixels in an image can also be calculated using a macro ImageSize multiplies the x y and zd imensions of the image and returns the result Programming with Image Image 2 1 Library User manual 7 The maximum is calculated for every pixel in the images A while loop is used which terminates when all pixels in the image are processed Within the loop the corresponding pixels of the two input images are compared and the one with the highest value is stored in a variable 8 The maximum value of the two pixels is stored at the corresponding location in the output image 9 After a pixel has been processed each pointer to the data is increased by one to point to the next pixel in the image This continues until npix pixels have been processed 10 After proce
190. ion Programming with Image 9 19 Image 2 1 Library User manual include image h 1 int maximum IMAGE inl IMAGE in2 IMAGE out 2 PIXEL srcl src2 dest maxi 3 unsigned long npix if pre_op inl in2 COMPARE G 2D SPEC G_3D_SPEC OUT_AS_IN pre_op inl out ADJUST G_2D_SPEC G_3D_SPEC OUT_AS_IN 4 return NOT_OK srcl ImageInData inl address of first image 5 src2 ImageInData in2 address of second image dest ImageOutData out address of output image npix ImageSize inl number of pixels 6 while npix 7 if srcl gt src2 which value is higher maxi srcl else maxi src2 dest maxi store biggest 8 srcl t go to next pixel 9 src2t t dest return post_op out 10 Explanation of the function code 1 The necessary data structures and definitions are in the include file image h 2 The function takes three parameters which specify images two of which are input 3 4 5 6 9 20 images and one of which is an output image To access the data in the i control for the number of mages pointers are used A counter npix is declared to pixels processed Since three images are
191. ion counter with the actual operation counter of the image When they are equal it is assumed that the minimum and maximum value stored in the image info are still valid for the image and consequently these values are returned instead of recalculating the image range again 9 10 Programming with Image Image 2 1 Library User manual Image Color Lookup Tables The clut field in the IMAGE structure can be used to attach a color lookup table CLUT to an image Although most commonly used for influencing only the display of an image some images palette images cannot be shown or interpreted without one For now in Image the color lookup tables in Image are fixed length RGB color model only typedef struct clut_t char clut_name IM_NAMELEN unsigned char r 256 unsigned char g 256 unsigned char b 256 unsigned long table 256 t CLUT clut_name a string used to identify this lookup table The maximum length of the string is IM_NAMELEN see image h characters trailing 0 included r g b arrays containing the RGB triplets specifying the colors table array that is reserved for use by the display interface attached The interface can use this array to map the RGB triplets of ther g and b array to the windowing system s colormap The Image library does neither fill nor interpret the contents of this array in any way After initialization the Image library supplies a number of default color
192. ion with the input data and the latter with the output image data for each line in the image The data in the common_line is stored either in long words or in doubles The choice of which is determined by the routine that stores a line of the source image into the common_line The receiving routine of the destination image must therefore be able to read the data from the long word format and from the double format The receiving routine does the actual conversion To summarize every image thus needs two routines for this conversion e One to store the data of the image on a line by line basis into a common_line structure either in long words or in doubles e One to read the data from the common_line and store it in the output image This routine must be able to read the data in the common_line structure which means that it must be able to read both long words and doubles COMMON _LINE structure The common_line structure COMMON_LINE is defined in the file image h like this define COM_LONG 21 define COM DOUBLE 22 typedef struct common line t int type type of data void data pointer to the data unsigned int x Y Z dimensions and position line unsigned int t time position of the line unsigned int nr_ channel number of channels per pixel double hint_min minimum value in data 9 32 Programming with Image Image 2 1 Library User manual double h
193. irectory contained in the MACRO environment variable see Chapter 2 for instance in the scilimage14 prog example directory If you do not do this SCIL_Image cannot find your function You can not use this function using the File Run command that is reserved for complete C programs which should contain a main To use the function exit the editor by clicking on the close box of the myfunc c window upper left button Dmot use File Exit SCIL to quit it will quit SCIL_Image altogether You will automatically return to SCIL_Image From SCIL_Image you can now use the function myfunc by typing in its name and list of parameters myfunc cermet a 127 When the Enter key is hit your function will be loaded and executed immediately You can not abbreviate your function to a SCIL_Image style commandmyfunc cermet a 127 because that could only be done if the function would have an associated CDF file see chapter We will demonstrate below how to do that In Making a New Compiled Version of SCIL_Image we show how you can incorporate your function into SCIL_Image as a full fledged compiled command To help you manipulate your own programs in SCIL_Image there are some commands you may find useful They can be found under the SCIL menu but may also be typed in We discuss some of them now If for example you wish to remove the current program from SCIL_Image s memory use the
194. is represented in the interpreter as an int the TRANSLATE line in the CDF file therefore reads TRANSLATE odd int The translate entry must be located in the CDF file before the cdf type it describes is actually used in the description of a command variable entry The lt variable entry gt contains global variable descriptions one for each global compiled variable Such a variable description is used by SCIL_Image to access these varaibles from within the interpreter The syntax for a variable description is lt variable entry gt VAR name type VAR name type A variable entry starts with the key word VAR followed by the name of the global compiled variable Advanced SCIL_Image 5 5 SCIL_Image 1 4 User Manual The type of the variable may be specified surrounded by string delimiters but this is not required If no type is specified the variable is assumed to be an int Valid types are the standard C types and pointers to them Structures and unions must be completely defined in the type string It is not possible to refer to defined structures like FILE or IMAGE command entry The lt command entry gt contains command descriptions one for each command Such a command description is used by SCIL_Image to build a dialog box for the command and to check arguments defaults and the expansion of names The syntax for a command description is lt command entry gt FUNC name
195. is called an overload file its syntax is explained in the chapter Programming with Image You only need to make an overload file for commands that should be able to handle various image types To make the command add_one usable under SCIL_Image and bring it under the menu we make a file in scilimage14 prog comfiles This file is similar to the CDF file we saw in Sample Session Three add_one cdf function description FUNC add_one ENU Arithmetic image A Input Image image B Output Image It determines the default values of the images and helps to pop up the appropriate dialog boxes We now have all ingredients to make a new version of SCIL_Image that contains our add_one command We still have to compile the file add_one c and to make a new version of SCIL_Image 1 Open the Microsoft Developer Studio workspace c scilimage14 prog scilimag dsw if not already open and add the file add_one to the project mylib right click mylib and choose Add files to project and then select the file add_one c from the directory you put it in 2 Open de file c scilimage14 prog comfiles mak set the Open as to text in the dialog box and at the end of the COMFILES list add c scilimage14 prog example add_one cdf and save the file fill in the appropriate path name if you saved the file add_one cdf somewhere else 3 Open de file c scilimage14 prog overload mak set th
196. is mechanism to create a complete separation between the image processing and the user interface Consequently Image can be used with any user interface or none at all A Graphical User Interface GUD can use this publish and subscribe mechanism to create all kind of independent interface objects that react only to the object s to which they have subscribed Adding new interface objects can then be done without the need to change the source code of the existing objects It can even be done without having access to this code For example A graphical user interface GUI will contain an image viewer object that is visualizing an image on the screen When the image data changes and the image publishes this message the viewer is notified and may act by visualizing the changed data Just as easy a histogram data viewer can subscribe itself to the image and visualize the histogram of the image These viewers do not know of each other s existence and do not have to just as the image does not need to know that two viewers are displaying its data In a later stage without the image knowing it the viewers can unsubscribe themselves from the image and may attach themselves to another image Object requirements For an object to publish events and allow other objects to subscribe to it it the structure must contain a void publish field as its first member The mechanism uses this field to store information regarding the subscriber
197. ist if edge_object C object area object lt 20 Il area object gt 400 hide_object C object Erase object from image rm_object object Mark object for removal from list o_list update o_list Update the object list Next the optical density of the remaining objects is measured FORALL object o_list object_dens_meas A C object OD To store the grey value image of the objects in image A with a mean optical density lower than 50 an image silo must be created long silo silo create_silo test silo int object_number 0 FORALL object o_list if od_mean object lt 50 object_rect_to_silo silo object_number A object close_silo silo Analysis of Images and Objects AIO 10 5 Image 2 1 Library User manual In the next example we show how objects stored in an image silo can be positioned in an image silo open_silo test silo int num num get_free_entry silo Get number of objects in silo for object_number 0 object_number lt num object_number part_from_silo silo object_number D 40 40 Finally we show how an image silo can be presented in the form of a composite image Note that although not shown here individual objects can be added to a composite image as well long comp clear_im D comp start_comp D silo_to_comp silo comp 1 100 Note that we have shown the examples as C code fragments In the SCIL_Image package these
198. l The IMAGE structure All images are referenced through a pointer to an IMAGE structure regardless of the type of the image The actual data of an image in Image is stored in a contiguous piece of memory An image pointer is type omnipotent in the sense that the referenced image may change type but the image pointer is retained The IMAGE structure consists of two parts The first part contains information that is for all image types One of its members points to the second part which is image type dependent The second part is called the image type descriptor The general part of the IMAGE structure typedef struct image_t void publish Publishing hook int type_ident Type Identifier long type_spec Data structure specification int lenx Image Width int leny Image Height int lenz Image Depth int slice Slice that is displayed 3 d only int o_lenx Old Image Width int o_leny Old Image Height int o_lenz Old Image Depth char image_name IM_NAMELEN Image Name char window_name IM_NAMELEN Display Window Name void clut Image color lookup table void graphics Image Graphics Specific Info ROI roi Region of Interest Descriptor void in_descript Image Type Descriptor input void out_descript Image Type Descriptor output long flags Image Flags int op_cnt Operation Counter void image_info Image Info List void future4
199. l image analysis with a focus on objects and attention to the storage method for images of individual objects and the human interface Components of the AIO framework The AIO package is completely based on linked list manipulations and computations on individual objects The framework consists of various parts which we now consider briefly Labeling objects During the image labeling process a data structure is assigned to each image object to store information and measurement results as they become available To allow multiple objects in an image to be viewed as a unit and to support image measurement a linked list of objects is built with information regarding each object s dimensions and positions Measuring individual objects For feature extraction each object is passed to a measuring routine which calculates the specified features storing the result in a feature value pair connected to the object data structure Measurements are carried out in sub images processing only pixels in the rectangle enclosing the object The time needed for measurement is therefore decreased substantially Further because the features need not be specified simultaneously a pre selection based on simple operations can be made for object classification The fact that features to be measured need not be specified simultaneously is highly useful for purposes of object classification A Analysis of Images and Objects AIO 10 3 Image
200. l be logged Offset lt value of offset gt must be positive pre_op checks the images and adjusts the output image to match the input image see The pre_op function on page 9 14 If for whatever reason pre_op detects that something is wrong it will return with the value 0 and the operation will be stopped by the return im_report_error statement If all checks are passed successfully the operation will perform its calculations followed by a call to post_op for the image out Check functions The available check functions are image_ok checks if the pointer supplied is a valid image pointer 9 24 Programming with Image Image 2 1 Library User manual images_ok checks if the two pointers supplied are valid images image_readwrite_ok checks if an image is writable no read only flag clut_ok checks if the pointer supplied is a valid color lookup table pl_io_ok checks if the specified bitplanes are in the correct range plane_ok checks if the plane is in the correct range im_val_ok checks if value s is are smaller than the image size s val_check checks if a value is smaller than the image size iter_ok checks if the number of iterations is positive or zero edge_ok checks if the edge bit parameter is correct con_ok checks if the connectivity parameter is correct con6_ok checks connectivity parameter on 4 8 48 or 84 odd_fsizes_ok checks if filter sizes are odd and in the specified range range_ok check
201. l interactive processing environment SCIL_Image has been developed as a multi level environment Figure 1 3 shows the internal levels of the environment SCIL Library handler To provide expansion of the system software libraries are linked via the library handler The system comes with quite a few libraries for image processing functions Through the Library handler the system can be expanded with external libraries or with any other user provided software library The data structures to get access to the image or parameter data as in use in SCIL_Image are documented in Programming with Image Compiled routines can be integrated in the system by adding an entry to a Command Description File 1 10 Outlines of SCIL_Image SCIL_Image 1 4 User Manual Command SCIL menu and Description dialogue system File SCIL command expander SCIL C Interpreter SCIL library handler Figure 1 3 The multi level interactive environment SCIL SCIL C interpreter Using SCIL a C interpreter for command interpretation ensures general programmability since algorithms are developed with the full capabilities and flexibility of the C programming language An additional advantage is that because development takes place in an interpreter using standard C syntax the source code can be compiled and incorporated in the system with the use of the library handler SCIL Command expander The basic task of the command expander is to tra
202. ld in the ROI structure is used to define the shape The BOOL_MASK structure see roi h is defined as typedef struct bool_mask_t BYTE data int lenx int leny int lenz BOOL_MASK Every time an arbitrary shaped ROI is used the data in the ROI is copied to a separate image like rectangular ROIs using a mask to distinguish the pixels that belong to the ROI from the other pixels This mask is a Boolean mask which is similar to a binary image in the sense that all elements have a value of 0 or 1 A mask can be made by converting a BINARY_2D or BINARY_3D image into a BOOL_MASK using the function get__bool_mask This function returns a pointer to a BOOL_MASK that can then be passed the function roi_define BOOL_MASK is a separate type in Image for several reasons First binary images contain a lot of data not needed for the mask Furthermore it is necessary that the mask is present and valid for as long as the ROI exists Using a binary image will cause problems whenever this image is unintentionally used in another operation The mask structure is unlikely to be used accidentally in other operations It should be noted that when an arbitrary shaped ROI is used the ROI image is still rectangular and the pixels in the ROI image that where not under the mask are set to 0 The operation performed on the ROI does not know whether it is an arbitrary shaped ROI or not and thus will operate on all pixels in the r
203. ld the image at a given level and then show the contour in a display window We call this program myfunc We would like to give it an file name an image and a threshold value as arguments Make the new file myfunc c by selecting File New C Program from the menu Type in the contents of the box below include image h myfunc filename image level char filename IMAGE image int level readfile filename image 0 0 threshold image image level contour image image 0 8 0 Let us use this opportunity to show you some features of SCIL_Image that are very convenient when you are developing programs Suppose that you had forgotten exactly what the arguments were of the function readfile You can easily pop up the required information without leaving the editor You simply select the text readfile with the mouse and type Ctrl Enter The dialog box pops up and by clicking the Help button you obtain the Help page on the readfile command Note that in the dialog box you can also see what the default values of the parameters to this function are Select File Save from the menu and e give it the name of the main function in your program followed by c in this case myfunc c The name of your file shoulda ways be the name of the main function defined in it followed by c 3 26 Getting started SCIL_Image 1 4 User Manual e store it in a d
204. lines describing the contents Definition of the image type this table is for message functions pair definition see below message functions pair definition message functions pair definition message functions pair definition The rules defining this layout are e All lines that start with a pound sign are comments Empty lines are also regarded as comment e The first non comment line found in the file must be a table definition using the following syntax TABLE lt window name gt lt IM IDENT gt lt IM SPEC gt lt IM DIMENSIONS gt lt window name gt one of g2d g3d f2d 3d b2d b3d c2d c3d col2d col3d 12d 13d lt IM IDENT is the type identifier for this image type lt IM SPEC gt is the type specifier for the image type 9 30 Programming with Image Image 2 1 Library User manual lt IM DIMENSIONSS is the number of dimensions the image type has The standard table for the GREY_2D image type is specified by TABLE g2d GREY_2D G_2D_SPEC 2 e The lines in the file following the line with the TABLE keyword are message function pairs The first word on the line is the name of the operation the generic function layer and the second word is the name of the function that performs that operation for the image type involved When overload c is created the first word is converted to a string and the second one into a function address For the add_im operation the message function
205. lines in the image that have not yet been transferred the function should return the value 1 or any other non zero value When the last line is processed the function returns the value 0 zero The source of the default grey value function is listed below for reference 9 34 int g_2d_conv_to_common IMAGE im int n COMMON_LINE com_line GREY_2D_IMAGE g_im PIXEL pix long data register long npix im_begin_func g_2d_conv_to_common g_im GREY_2D_IMAGE ImageIn im npix Grey2dImageWidth g_im if n 0 First time called data long malloc size_t npix sizeof long if data return im_report_error g_2d_conv_to_common TE_NOMEM no memory allocated for data set_common_line com_line COM_LONG data 0 0 0 0 1 0 0 255 0 4 data com_line gt data com_line gt x npix com_line gt y n com_line gt z Os pix Grey2dImageData g_im pix com_line gt y npix Programming with Image Image 2 1 Library User manual while npix gt 0 datat pixt if n Grey2dImageHeight g_im 1 im_end_func g_2d_conv_to_common return 0 0 stop this is the last line im_end_func g_2d_conv_to_common return 1 1 more lines are available Destination function specification The header of the destination function is specified as follows
206. lter ARGS image in A Input Image image out B 5 Output Image switch con sum sqrt sum Sum OR Sqrt of quad choice is used for variables that select among a number of pre defined choices Each choice can have an argument value which is passed to the function when the choice is selected Often this is an image identifier or an integer The choices are listed separated by bars in the min field The max field is empty and should therefore contain a minus sign An example Advanced SCIL_Image 5 11 SCIL_Image 1 4 User Manual im_aio cdf UNC aio_label ANUAL aio_labl ingle Objects Conversion image in B Input Image image out C Output Image choice con 8 48 Connectivity A choice may have an alias associated with it In this way SCIL_Image can prompt you to select among meaningful terms rather than among integers Suppose for example the valid choices for a particular argument are 1 2 and 3 but that they actually are choices of the X Y or Z axis You can make this explicit by substituting 11213 in the min field with X 1 IY 2 IZ 3 In the dialog box and at the command line you will be offered a choice among X Y and Z rather than among 1 2 and 3 An example image cdf UNC contour ENU Special_Points ARGS image in B Input Binary Image image out B Output Binary Image switch in Off On Off Set Edge Pixels choice in 8 4 8 Connectivity c
207. lue of one bit in a bitplane the CPU has to fetch all the bits of one pixel typically 8 or 16 All irrelevant bits 7 out of 8 or 15 out of 16 have to be masked out Using a 32 bit computer just to fetch 1 relevant bit at a time is a bit overdone A second disadvantage is that even if only 1 binary image is needed still the memory for 8 or 16 bitplanes has to be allocated In Image a bit mapped representation of binary images is used Each 32 bit addressable word contains the bit values of 32 horizontally consecutive pixels A 32 bit computer is optimally used in the sense that all 32 bits fetched in one memory cycle are relevant Not only 32 pixels in parallel are transported from and to memory but also the operations are done on 32 pixels in parallel Furthermore there is no need to allocate more memory than is actually needed to represent the binary image Figure 11 1 visualizes the differences between a pixel mapped and a bit mapped binary image 11 4 Bitmapped binary images Image 2 1 Library User manual BINARY IMAGE y j word containing X gt 32 binary pixels byte containing one binary pixel Figure 11 1 Pixelmapped versus Bitmapped Images All the 32 pixel words are stored in a linear array in the computer memory To facilitate the development of efficient algorithms we have chosen not only to store the pixels in the image itself but also a border around the actual
208. m increment_im decrement_im negation_im shift_im and_im orim orim invert_im eval clip Figure 3 24 the Arithmetic menu This menu contains various commands to perform arithmetic operations on images They are divided into three groups Variable arithmetic operations Int_based Exotic perform computations on SCIL_Image variables Image arithmetic operations add_im clip perform arithmetic computations on image pixels point wise for the complete image this allows you to divide one image by another image for instance Arbitrary evaluation consists of the very useful command eval which evaluates C like expressions with images and image coordinates Itools GreyMap Zoomer CTRL M Threshold lens stop lens cube_view Imeasure Figure 3 25 the Itools menu This menu contains some interactive tools We described them in Session One Viewing Images Getting started 3 41 SCIL_Image 1 4 User Manual The Properties of Text Windows In the sample sessions you have encountered various text windows and their properties Here is asummary There are 3 types of text windows in SCIL_Image They are the Worksheet the History window and any number of RunEdit windows Er Worksheet ioi x History me E I RunEdit Untitled1 The following key sequences can be given in any text window after selecting a command or a num
209. m The width of each cell already calculated and stored in the bin_width array can be calculated by the formula hbin_median lbin_median dth bin_ widt bins 1 The center pixel value of each cell n of the histogram can be calculated through center_ value lbin_ median n bin_ width In which n is the number of the cell starting from 0 To determine in which cell certain pixel value would be stored the following formulas are used bi idth low _ border lbin_median a 2 0 _ pixelvalue low_ border bin_ width bin For instance the histogram of grey valued integer image data ranging from 0 to 255 and stored in a histogram of 256 cells would show the following values lbin_median is 0 the lowest value of the image data hbin_median is 255 the highest value of the image data and bin_width is 1 each cell is exactly 1 pixel value wide Consider the histogram of floating point data ranging from 0 0 to 1 0 and stored in 100 bins The Ibin_median is 0 0 the hbin_median 1 0 and the bin_width 1 0 99 0 0101 The center pixel value of bin 62 would be 0 6262 0 0 62 0 0101 The histo_data function calculates the histograms of all image types in Image except for complex images This function is capable of calculating the histogram of sub ranges of the image data When such a histogram is calculated the above formulas remain valid but after the bin has been calculated it must be checked if it is
210. mands By the menu system help can be offered to the user as well as default values of parameters can be suggested The menu system provides a user friendly interface suitable for novice users Based on the above discussion we believe an image processing environment should fulfill the following requirements e The system should contain an extensive set of image processing routines readily available to the user e Visual feedback must be given in immediate response to selected or specified commands at least the ones which modify the image e Amenu system and command interpreter should be available simultaneously to support novice and advanced users Outlines of SCIL_Image 1 9 SCIL_Image 1 4 User Manual e It should be possible to both develop and use applications within the same environment to escape the gap between program development and use e One package should be made available on a variety of machines rather than separated ones developed for each machine type This includes the window system and the command language e Jt should be possible to execute commands specified in a shorthand form e The image processing functionality of the system should be extensible both with private routines as well as routines from other packages accessible as a subroutine library As an example consider extending SCIL_Image with an interface to a database package a spreadsheet or a statistical package A multi leve
211. me that you want your function my func to appear as the command myfunc in the standard Options menu of SCIL_Image and in a new menu which you want to call pene Then you should edit myfunc cdf so that it contains Ww lyMenu SSCILIMAGE FUNC myfunc MENU MyMenu Options ial NOT_COMP TILED filenam trui My File image A My Image odd 127 1 255 Only odd values Make sure that no empty lines follow the last line of myfunc cdf In order to add myfunc cdf to comfile 1 Place myfunc c in one of the directories listed in the environment variable MACRO Chapter 2 such as scilimage14 prog example myfunc c 3 28 Getting started l a ______ SCIL_Image 1 4 User Manual 2 Store the CDF file myfunc cdf in a directory of your choice for instance scilimage14 prog example myfunc cdf 3 Exit SCIL_Image if it is running and start the Microsoft Developer Studio devstudio by double clicking the file the workspace file for SCIL_Image c scilimage14 prog scilimag dsw 4 From within the Developer Studio op the file c scilimage14 prog comfiles mak be sure to set the Open as option to Text otherwise the wrong things happen At the end of the COMFILES list add the line SCIL_ROOT prog example myfunc cdf make sure to add it onto the last line in the block COMFILES and preceding the Return on the previous line by a backslash just
212. menu system always prompts you but the command line system does not In that case you can just type a after the command and it will ask you for all the parameters that it needs type readf The system prompts you for the arguments of the readfile command one by one You can either press the return key when you like the default value or enter your desired value You can also pop up the dialog box of a command by selecting the command in the Worksheet and hitting Ctrl Enter either of the Enter keys will work type readf select it by double clicking on it and hit Ctrl Enter An important feature of SCIL_Image is that C control statements can be mixed with image processing commands such as in int i declaration of a variable for i 0 i lt 200 i i 10 thresh A B i As you see this creates a for loop which thresholds the image at different levels The variable 7 is now known to the system so if you want to see this again executing only the second line 3 22 Getting started SCIL_Image 1 4 User Manual is enough You need not re type this line just execute it again by selecting it and then clicking on Enter You can interrupt the execution of any command by using the Pause Break key Run the previous command again and interrupt it SCIL_Image is then ready for the next command The Worksheet in combination with the mouse offers a flexible way of giving comman
213. morphological operation is used Other examples show the same pattern In Image we provide a new algorithmic implementation of the basic binary morphological operations on general purpose sequential computers which are between 10 to 50 faster than existing fast implementations VWERWER GROEN In contrast to the previous fast implementations the provided algorithms do admit the use of structuring elements of arbitrary shape within a 65x65 neighborhood This is based on our belief that implementations of the basic morphological transforms should be valid for arbitrary structuring elements without strong performance degradation as non circular structuring elements prove to be very efficient in solving specific problems The speed of common implementations of morphological image transform in most cases if not all is linear proportional to the number of pixels in the structuring element We use logarithmic decomposition of structuring elements to efficiently process large convex structuring elements Compared with the standard linear decomposition logarithmic decomposition results in much faster algorithms Erosions Dilations and Logarithmic Decomposition The basic image operations defined in mathematical morphology are the erosion and dilation Let A k l be a binary image for 0 lt k lt K 0 lt l lt L By definition we say that a pixel k l is an object pixel if A k 1 If A k 0 the pixel k l belongs to the backgrou
214. mple session we will show you how to embed your functions as commands in a new compiled version of SCIL_Image We take a simple example in image processing At first this example shows you how to obtain pointers to images then we extend it to use the convenient error checking functions of SCIL_Image Finally we extend it once more to make one command that treats images of various types in this case integer valued and real valued images We will write a program for a command that adds 1 to an image not very useful but it illustrates the issues involved We enter it using the File New C Program menu selection add_one c first version include image h g_add_one IMAGE in IMAGE out PIXEL indata outdata long num indata ImageInData in outdata ImageOutData out num ImageSize in while num outdatat indatat 1 display_image out Note that we do not need to start with main since SCIL_Image will be the main program that will call this function Thus the terms program and function are almost synonymous We use program for the contents of the file including the include statement and possible other functions A function is then any of the function bodies in the program The main function is the function lt any_name gt defined in the file lt any_name gt c The program itself looks much like you would expect from a C prog
215. n both viewports The use of several viewports on the same image can be combined with the tool buttons on the status bar You select viewports with the right mouse button whereas you select windows with the left button select one of the viewports on A by clicking on it with the right mouse button A green border in the viewport indicates that it has been selected click the zoom and pan icon os y A Zoom amp Pan interaction window pops up with a miniature viewport displaying image A BE H Help Figure 3 6 The Zoom amp Pan window When you point and drag on the little copy of A in the Zoom amp Pan window the selected viewport changes as if you were scrolling it When you click on one of the numbered buttons the zoom factor of the display changes you may also change this more continuously with the sliding bar play around with the Zoom amp Pan options Again a change in image A is reflected in all viewports You may verify this by reading in the original image select Image I O readfile click Browse double click c scilimage14 images schema dat click OK 3 8 Getting started SCIL_Image 1 4 User Manual All three viewports on image A change their displayed contents Zoom amp pan can handle several images at the same time To show how this might be useful we do some simple processing on image A we threshold the image that is we make a binary two valued
216. n dealing with numerical non image data the var_objects described in Var_objects on page 9 37 can be useful Using HISTOGRAM objects which can hold multi dimensional and arbitrary length histograms is discussed in Histogram objects on page 9 40 9 2 Programming with Image Image 2 1 Library User manual Image infrastructure The infrastructure is designed to be independent of specific image types in order to provide a consistent framework which can support a growing variety of image types Image types can be divided into classes and subclasses For this purpose two labels are associated with each image type These labels can be found in the IMAGE structure see The IMAGE structure pg 9 6 The first is called type_ident type identifier and has a unique value for each image type The second is called type_spec type specifier and specifies the data representation and layout of the image type The number of classes is limited to 32 The number of subclasses is unlimited The image types their type idents and type specs which are currently implemented are image type type_ident value type_spec _value grey valued 2 d GREY_2D 1 G_2D_SPEC 1 binary 2 d BINARY_2D 2 B_2D_SPEC 2 floating point 2 d FLOAT _2D 3 F_2D_SPEC 4 complex 2 d COMPLEX_2D_ 4 C_2D_SPEC 8 grey valued 3 d GREY_3D 5 G_3D_SPEC 16 binary 3 d BINARY_3D 6 B_3D_SPEC 32 floating point 3 d FLOAT _3D 7 F_3D_SPEC 64 complex 3 d COMPLEX_
217. n felt desirable to provide the user access to only those menu items which are meaningful for that particular applications To accomplish this the user interface of the SCIL_Image system can be altered at will by adding removing and re ordering menus and menu items SCIL and the Image library Although from the outside SCIL_Image looks like a fully integrated environment in reality it is constructed from two separate entities SCIL and Image SCIL being the C interpreter and menu amp dialog generator And Image the Image 2 1 library being the Image infrastructure and a large set of image processing functions These two parts supplemented by a user interface for the visualization of the images make up the SCIL_Image package The Image library can be used separately to create stand alone end user applications This means that after trying out some new ideas and gradually developing image processing functions in the full featured SCIL_Image environment they can be used at once in these stand alone applications Outlines of SCIL_Image 1 5 SCIL_Image 1 4 User Manual On reading this manual This is version 1 4 of the SCIL_Image manual To work with SCIL_Image it is not necessary to read the entire manual neither is it necessary to have experience with the programming language C even though the system is based on a C interpreter When using the SCIL_Image system for command execution only it is sufficient to kn
218. nction known to the system in the same way A large variety of filter and image manipulation routines are available These include arithmetical operations Fourier transformation geometrical routines and image measurement A number of these have been adapted from the well known TCL Image package There are too many functions to discuss here we only give a few examples A quick overview is given in the guided tour in Getting started chapter 3 TCT Bitmapped functions for AIO library for fast mathematical interactive object morphology with arbitrary measurement structuring elements i Chapter 10 Chapter 11 The Image infrastructure and display Image processing operations all separated per image type Chapter 6 and 7 Chapters 9 and 12 Figure 1 4 The Image libraries Image infrastructure Image is based on a division of the images and image processing routines by type This type mechanism simplifies the management of images and makes programming with different types of images easy Each type has its own functions for type dependent tasks that are accessed by function overloading Currently available types in both 2D and 3D are binary 1 12 Outlines of SCIL_Image SCIL_Image 1 4 User Manual grey value float complex label and color The image infrastructure is particularly helpful in maintaining a fail safe program when programming in an extensive image processing system When you are an advanced programmer of S
219. nction will then clean up the necessary intermediate data structures In order to handle the conversions the image type descriptors in the IMAGE data structure are used Consider again the example of thresholding an image in place The pre_op 9 12 Programming with Image Image 2 1 Library User manual routine creates a binary image and connects it to the output image descriptor out_descript When the post_op function is called it destroys the data pointed to by the input image descriptor in_descript and the input image descriptor itself It then connects the output image descriptor to the in_descript field of the IMAGE structure and displays the new image data The situation pointed out above looks like this Before the call to pre_op the IMAGE structure has one descriptor attached to both in_descript and out_descript IMAGE structure type descriptor type GREY_2D Figure 9 1 IMAGE structure layout before pre_op out_descript Since the image had to change to type binary and yet remain grey value for input pre_op split up the in_descript and out_descript and attached different descriptors to them The image now holds both the grey original and the binary destination IMAGE structure type descriptor type GREY_2D type descriptor type BINARY_2D the grey image Figure 9 2 IMAGE structure layout after pre_op out_descript After the actual proce
220. nd A structuring element is also a binary image most often a small one The erosion of the image A with structuring element S resulting in the image B is the neighborhood operation defined as 11 2 Bitmapped binary images Image 2 1 Library User manual B k l ANDs ij 1 Alk l j 1 i e we take the logical AND of all the pixels in the neighborhood of k l as defined by the structuring element S The above definition states that a pixel k l is an object pixel in the erosion result only if all pixels in the neighborhood defined by the structuring element S centered at k l are object pixels in the original image A The dilation of the image A with structuring element S is defined as Bik ORSfijj 1 Alk il j 2 i e instead of AND ing all pixels as is done in the erosion we take the logical OR of the pixel values The computational complexity of the erosion and dilation is linear proportional to the number of pixels in the structuring element For a square structuring element of NxN pixels the time complexity thus is quadratic in N For large N the decrease in speed therefore is dramatic Simple but often used structuring elements however can be decomposed into smaller ones and so the time complexity can be reduced In this section we will only use the 4 connected neighborhood This structuring element will be denoted as D To illustrate 3 different approaches to decompose convex structuring elements consider t
221. nd e Fills in missing arguments with default values e Executes the C code and displays the results The first two steps are called command expansion This feature of SCIL_Image means that you do not need to use the exact name or arguments of the C codes that do image processing For instance you can just type readf trui without knowing that this is actually the C statement readfile c scilimage14 images trui A 300 300 SCIL_Image is also capable of interpreting C statements This means that if you want to execute a command 10 times you can type for i 0 i lt 10 i lt command gt because SCIL_Image understands the control flow of standard C A sequence of commands to SCIL_Image may therefore look very much like C code You can collect such sequences in files and execute them Such files are called macros SCIL_Image can also understand actual C programs and run them without compiling We call these programs interpreted programs In this chapter we explain these three ways in which SCIL_Image interprets C as well as some basic commands that help you develop your programs However a warning before you start SCIL_Image s C language is close to Kernighan amp Ritchie 1978 C but there are some limitations see Features of SCIL_Image s C interpretei on page 4 16 Kernighan B W Ritchie D M The C Programming Language PRENTICE HALL INC Englewood Cliffs New Yersey ISBN 0 13 110163 3 4 2 The C I
222. nd output descriptor of the specified image are different If they are it throws away the image that is connected to the input descriptor in_descript will then be set to the same descriptor as out_descript which is the result of the calculation Finally it will publish that the image is changed which is a signal for a Graphical User Interface to redisplay the image ROI and post_op If the output of an operation is a ROI then pre_op has already checked whether the result would fit in the ROI post_op then copies back the result to the parent image of the ROI and publishes that both the ROI and the parent image changed If the ROI is of a different type than the parent post__op converts the parent to the same type as the ROI and thus destroys the previous contents of the parent This is done because the parent was implicitly specified as being the output image by specifying the ROI as output A simple example This section contains an example operation which calculates the maximum value of pixels at corresponding locations in two images and stores it in another image The output image in the example is in this case of the same type as the input images Please note that the example is restricted to implementing the maximum function for the image types of GREY_2D and GREY_3D only If a routine should operate on more image types use of the function overloading facilities page 9 27 is strongly recommended The maximum operat
223. nd will therefore be used as an example First some rules and guide lines New image types 12 19 Image 1 2 3 4 2 1 Library User manual The sub type must have an unique t ype_ident The value of the type_idents of the standard image types are listed in the header file image h The values 1 to 32 are reserved for main types and should not be used for a subtypes See the difference between LABEL_2D and GREY_2D GREY_2D being a main type is in the range 1 to 32 and the subtype LABEL_2D is 33 The t ype_spec of the subtype must have the same value as the t ype_spec of the main type A new define is recommended for readability of the source code In image h thet ype_spec L_2D_SPEC of the label_2d image type is defined to be equal to G_2D_SPEC the type_spec of grey_2d In order to make use of the functions of the main type the type descriptor should be the same as that of the main type The subtype can then use the functions from the existing image type In the header file label_2d h the type_descriptor LABEL_2D_IMAGE is defined as the type descriptor GREY_2D_IMAGE A new name and new macros are defined to be used in the source code for this subtype If additional fields in the type descriptor are necessary copy the structure of the main type and add the extra fields at the end It should be realized that the functions that are inherited from the main type do not know about these fields and therefore
224. nds of text windows These are the Worksheet the History window and the RunEdit window which you create when you choose File New C Getting started 3 23 SCIL_Image 1 4 User Manual Program see below In any of these windows you can select text with the mouse and if you then hit the Enter key SCIL_Image will attempt to execute the selection If you perform in any text window the actions select a command name by mouse hit Ctrl Enter in the dialog box which pops up select Help then SCIL_Image will pop up a Help page with information on the command SCIL_Image 1 4 Windows Manual Iof x File Edit Bookmark Options Help threshold NAME threshold thresholding into binary image bi_threshold thresholding with two levels into binary image SYNOPSIS include im proto h int threshold IMAGE in IMAGE out int level int bi_threshold IMAGE in IMAGE out int low int high DESCRIPTION threshold performs a thresholding operation on the grey value image in and stores the result in the binary image out Ifthe value of a pixel is greater than or equal to level the corresponding bit in the out image is set to 1 Otherwise it is set to 0 bi_threshold converts all the pixel in the range from low to high into 1 pixels in the output and all pixels that are outside that range to 0 pixels low and high are converted to 1 pixels RETURN VALUES IE_OK 1 on suc
225. needed The infrastructure can destroy the data when the image is destroyed when required The functions AddImageInfo GetImageInfo and RemoveImageInfo must be used to add the data pointer to and retrieve and remove the data pointer from the image The IMAGE_INFO structure typedef struct image_info char infoname IM_NAMELEN unsigned long info_size void dfunc void void infoptr IMAGE_INFO infoname a string used to identify the data attached to this structure The maximum length of the string is IM_NAMELEN see image h characters trailing 0 included Image reserves all names that start with SI_ for private usage info_size size of the data attached not yet used dfunc pointer to a user application specific function to destroy the data when the image is destroyed This function is called with infoptr as the parameter When the data cannot be destroyed global variable or should not be destroyed shared with other images a NULL pointer should be specified infoptr pointer to the data The data pointer to is not referenced of interpreted by Image in any way Any type of data can be stored Image uses this mechanism to store the actual range of image data Whenever the range of the image data needs to be calculated the get_pixel_range function is called it checks if the range data is still valid by comparing the recorded operat
226. nments range from a collection of library routines via command driven systems or menu driven systems to highly specialized programming languages In the first category of systems each of the routines comprise a command No environmental support is provided Although this category offers the advantage of development in a standard programming language these systems are less suited for interactive image processing and are in general not user friendly Command driven interactive systems or command interpreters are characterized by prompt execution of image operations The functionality of command interpreters ranges from those which offer the execution of operations and the control of parameters to those which provide variable declaration expression evaluation flow control and or procedure execution It is important that command interpreters provide tools for the constructions of new commands composed of a sequence of basic commands Although systems in this category are geared towards interactive processing most do not have the flexibility and complexity inherent in a standard programming language Moreover the diversity of semantics and syntax between interpreters obstructs user acceptance An additional disadvantage of command interpreters is that the names of commands and the sequence and meaning of the parameters must in most cases be known by heart In contrast menu based interactive systems allow easy access to image processing com
227. nslate shorthand commands entered by the user to a complete C statement This is to avoid the user being bothered with the precise typing of the C punctuation whenever possible Simultaneously the command expander checks the arguments of functions for consistency and if any are missing fills in default values The command expander operates according to a Command Description File CDF in which the information on each command and its arguments is stored Outlines of SCIL_Image 1 11 SCIL_Image 1 4 User Manual SCIL Menu and dialog generator The menu and dialog generator provides an interface that makes it possible to select commands via the mouse rather than by typing them The interface is generated when SCIL_Image is started based on specifications in the Command Description File When the user selects a menu option a dialog box is generated The box gives the parameters of the command with their defaults and ranges The graphical presentation of the items in the dialog box is adjusted by the type and range of the parameters In SCIL_Image the menu and dialog generator performs an educated guess to produce the most convenient dialog The Image libraries The Image libraries offer a large variety of image processing tools Figure 1 4 gives an overview of the libraries available in Image As a user you will not notice the existence of different libraries because the library handler and the Command Description File make every fu
228. nt objects are displayed using different colors Since only eight colors are used for displaying labeled images the same color will be used more than once However separate objects of the same color have a distinct object id check with the left mouse button To show what kind of measurements can be performed on labeled images try the operation measure The command has to be selected through the menu for clearness 6 16 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual select Single_Objects measure choose Interaction Yes hit DO IT A new image called labelled_image pops up in the middle of the screen which contains the same data as image C To see this remove the dialogue box hit gt CANCEL Press the left mouse button while pointing at an object in labelled_image several numbers will be printed in the command window These numbers are the result of several measurements performed on the object Stop the measurement by placing the pointer in the display window or the command window and type lt RETURN gt Color images COLOR_2D amp COLOR_3D In this version of SCIL_Image version 14 color images can be represented by one of five different color models The present models are RGB CIE XYZ CIE Lab HSI and CMYK Data representation of color images The color models of color images are each represented in a different manner RGB each pixel is made up of a Red Gr
229. nterpreter SCIL_Image 1 4 User Manual ANSI C compatibility Although the C language of SCIL_Image is not ANSI C we strongly recommend that ANSI C coding is used whenever compiled not interpreted code is added to SCIL_Image The function calling mechanism of the C interpreter has been changed to the ANSI C calling convention Consequently adding compiled K amp R C code may cause the arguments to be passed incorrectly to those functions The Direct Command Mode You are in the Direct Command Mode when you type commands into the Worksheet As you have seen in Sample Session Three page 3 21 you can use the Worksheet to give SCIL_Image commands like readf trui followed by hitting the Enter key on the numeric pad Commands like these are expanded internally to call the C function readfile This command has four arguments SCIL_Image fills in the default values to make your command readf trui into a proper function call You can see what these default values are by typing readf or by popping up a dialog box type readf select it by double clicking and hit Ctrl Enter Thus the proper C statement corresponding to readf trui is readfile trui A 300 300 All commands in SCIL_Image have these three versions SCIL_Image menu selection select Image I O readfile SCIL_Image command readf trui C function call readfile trui A 300 300 You should bear in mind the diffe
230. nual v D Index readfile D NAME De readfile read an image from file Da SYNOPSIS Re include im_proto h DD IMAGE readfile char filename IMAGE image int xpos int ypos DF DE DESCRIPTION Read the image stored in file filename and put it in image image H De USE_NAME a NULL pointer is specified as the image a new image is created at DH position xpos ypos with the same name as the file Ifan image is already present D with that name that image will be used ral file formats are supported see DJ below cach of which have an obligatory extension Ifa filename is supplied with no D extension the function will append the obligatory extensions one at the time to find K the file DL DM ICS format Two files per image are present the data file with the extension ids ON and the header file with the extension ics Do TIFF format The read function is capable of reading TIFF files according to the DP TIFF 6 0 specifications The file must have an extension that starts with Da sit The extensions used for finding a TIFF file are tif and ff DR JPEG format The file must have the jpg or jpeg extension Ds DT TCL format The file must have the dat extension Duw AIM format The data files of the AIM format must have the im extension Data Dy files for which no header file with the extension hd is present are Dx assumed 10 contain
231. ny specific computer operating system filing system compiler etc Because of this independence the file format is widely spread A large number of specifics of an image can be stored in a TIFF file all meant to enable an exact reproduction of the image to be made in a different place than it was made The exact description of the format is not the goal of this manual and thus omitted The recommended extension for this format is tif all extensions that start with these three characters are 6 22 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual accepted SCIL_Image is able to read and write TIFF images that comply with the TIFF 6 0 specifications The JPEG format The JPEG image file can be used for grey 2D and color 2D images in SCIL_Image The JPEG standard is designed as a compression standard for continuous tone still images The image data is compressed lossy upon storage meaning that no exact representation of the data can be made when the data is retrieved from file again The advantage of lossy compression is the large compression ratio that can be achieved The JPEG standard does include a method for lossless compression but this is not yet available The JPEG format supported by SCIL_Image is actually the JPEG File Interchange Format JPEG FIF also known as JFIF which is the standard JPEG format with a few additional requirements This may cause some JPEG files to be rejected
232. o Using the SCIL_Image Menu System The following sample session introduces various aspects of image processing using SCIL_Image through its menu and dialog system Not all topics are covered but you should get an impression of how to work with the system Let us read in an image select Image I O readfile click OK This a 256 256 grey value image as its label A g2D 256 256 indicates The grey values are integers in this case ranging between 0 and 255 You can point click and drag at the image and obtain a display of the grey values in the status bar at the lower left Let us inspect some of the grey value filtering operations in SCIL_Image A aplacian determines a local and rather coarse approximation to the second derivative of the image select Image Filter laplace click OK When you look at the values of this image you will see that they may be negative They are displayed clipped to 0 and 255 so 126 is just as dark as 0 This is the standard lookup table for the display of g2D type images You can check the pixel values by dragging the mouse in the image and reading the values on the status bar There are quite a number of image processing operations valid for image type g2D both the standard linear filters convolutions and some non linear filters such as percentile filters and the edge enhancing Kuwahara filter select Image Filter Kuwahara Parameters are pre set to do the ope
233. o need all commands you give are recorded faithfully in the History window and whenever you wish you can use an editor to extract the commands you want to keep and store them in a file However should you give a command that gets SCIL_Image stuck or makes it crash then you will have no such opportunity In that case logon would have saved all you did in a file so you could salvage what is valuable or trace your mistake Therefore if you are developing a complicated application usinglogon might be a safe precaution macro i v lt macrofile gt The macro command executes the lines in a macro file as if they were typed in one by one Therefore the lines of macro files are restricted to commands you would give in the Direct Command Mode The C Interpreter 4 1 SCIL_Image 1 4 User Manual i With the i option every line is executed interactively each individual command from the macro file will appear on the screen followed by the prompt y n q Entering y or pressing lt return gt will execute the command entering n will skip the command and q will exit the file and returns you to the SCIL_Image Direct Command Mode y With the v option every line is written to the Worksheet window as it is executed The lines are not written to the History window just the macro command itself is Please note the difference between the macro command and the chain command macro loads file that contains direct SCI
234. o pre_op will set an image to a specific size include image h include im_infra h set_cross_dim 100 50 4 pre_op in out ADJUST G_3D_SPEC FLOAT_3D The output image in this case will be a floating point image with the dimensions x 100 y 50 and z 4 Programming with Image 9 17 Image 2 1 Library User manual ROI and pre_op In addition to image size and type adjustment the pre_op and post_op routines together also handle region of interest processing When a ROI is defined a new IMAGE structure is reserved for it and processed as if it were an image The only difference is that a ROI does not have a display of its own When a ROI is specified as input pre_op will copy the data from the parent in the region of interest to the ROI image This way the region of interest can be processed by low level routines in the same fashion as other images Note that with the introduction of the operation counter for images see Operation Counter on page 9 9 the data of the ROI is only copied when the operation counter of the ROI image is lower than the one of the parent image the operation counter of a ROI can never exceed the counter of its parent This prevents unnecessary copying of the data Using the ROI as an output image is only possible if it does not have to change size because the ROI is nothing more than a part of another image Its size is one of its primary attributes If a call to pre_
235. oad mak set the Open as to text in the dialog box and at the end of the OVERLOAD list add c scilimage14 prog My_src noise ovl and save the file fill in the appropriate path name if you saved the file noise ovl somewhere else 4 For both the comfiles and overload project choose the Build command on the context menu when clicking the project with the right mouse button 5 Then rebuild SCIL_Image by choosing the Build command on the scilimage14 project 6 Start SCIL_Image again It now contains your function as you defined it available under menus you specified in noise cdf Please note that just must build the comfiles and overload projects by hand first because Developer Studio does not recompile the files sysfunc c and overload c when necessary Adding On line Manual Files to SCIL_Image Every function added to SCIL_Image may have anon line manual page This is a section in the file scilimage14 help user_def rtf which gives the basic text and defines links and keywords used to make it into a hypertext for the Help facility You can make such a page yourself and connect it to the standard SCIL_Image Help facility You should do this by following the instructions on the Help page User defined Help Pages It requires the editing in Rich Text Format of the file user_def rtf located in the help directory This file contains a sample manual page that you can use to creat
236. of performing the basic morphological image transforms using structuring elements of arbitrary size and shape In order to speed up morphological operations with respect to commonly used large convex structuring elements the logarithmic decomposition of structuring elements is used Do not read this chapter e If you are a beginning user who just wants to use the binary or morphology operations You should read the chapter e If you are interested in background information on implementational aspects of mathematical morphology e If you want to know more about bitmapped binary images e If you are developing binary image operations Image 2 1 Library User manual Introduction Mathematical morphology as introduced by Serra SERRA is nowadays an important component in almost any software package for image processing For an introduction to mathematical morphology we refer to tutorials MARAGOS HARALICK Morphological image transforms are used in many divers application areas in image processing and computer vision Most morphological transforms are constructed by elementary morphological operations such as erosion dilation and hit or miss transform The speedup of the elementary operations is therefore important in many applications As a typical example consider the program to analyze drawings of electronic schemes called schema located in the demo directory In this program more than 50 times an elementary
237. of the same dimensions and type as image in1 If all checks and adjustments succeeded the low level function that does the actual computing is called 1_g_add When 1_g_add completes its task the result is displayed with the post_op function Because add_im is a point operation the same add routine 1_g_add_ can be used for both two and three dimensional images Because neighborhood operations are always dimension dependent different low level routines have to be created Processing the data The function 1_g_add low level grey value add is the function which performs the actual calculations on the image data include image h include im_error h void 1_g_add IMAGE inl IMAGE in2 IMAGE out register PIXEL srcl src2 dst register int npix im_begin_func 1_g_add srcl ImageInData inl address of input data of inl src2 ImageInData in2 address of input data of in2 00 address of output data of out dst ImageOutData ou npix ImageSize inl number of pixels in image inl while npix gt 0 dstt 4 srclt src2t t im enq func 1_g_add return The type of data in grey valued images is PIXEL The addresses of the data in the images and the number of pixels in the image are retrieved using macros that reside in the include file Programming with Image 9 29 Image 2 1 Library
238. ommands of the SCIL_Image Menus ci ccce qssesse0seeeeceooscessscguveatcaosbacsadentevnenes 3 37 File senenin aon E A L E E va AEE EI ATA NaS 3 37 Editera tonn a e a aa aa e aas A aeai 3 37 SOU ee i eea ieee a ee ie a 3 38 MACE E E EEEE E Need sacs hiosdae dtu Syieytigeeateaahies nas osuadtund gees seas 3 38 Displayen a a S ERE sos tas seal aodaes E amp o EOR wana Ee ERES 3 40 OPONSE a Eaa eea e oa EERE Ee debe S ES 3 40 Arithmeti asp sheds enced rese Asi aS eE EK EEES Ee Eaa asiaa abate 3 41 MOG C SEE EE EEE E EE E E E E EE ETET 3 41 The Properties of Text Windows iss cccstesvessdcespacedaacacesecetasuseceatusedexessyebaa coanbecauasucedanes 3 42 On line maals oae a e e i Ie ete a N aE 3 43 Reference manual s i3ec cia yes lenge teats eerie enea ete ee el 3 43 Chapter 4 The C Interpreter sessssoocccssssocoecsssssccecessosoccessssooeesesssesoessee 4 1 SCH Imag and Coet e REEE ER A O AS EASE E 4 2 ANSE C compatibility censeret ese ata ee a e e esh 4 3 The Direct Command Mode 264 2422 SEA Ee SEE EA 4 3 Che Macro IVE ee dps cata fe Ee E dees ee ae ved asd tle a acelin E EERS 4 5 The Programming Mode Interpreted C functions and UFOS 0 0 eee eeeeeeeeeeeeeeee 4 7 Program Development Commands cee eeeeesceesseceseeeseeeescecaeceseeeseeesaeessaeenseenees 4 9 chain lt tilcename gt far Soc essai a aaa ease e ous Beau cesta i 4 10 list start Ci eee eevee tie sree casio Lean ao as oot ee Ales os ce ies els aC Si ew is 4
239. oncerning histogram This message is published just before the actual destruction of the object The destruction of the object can not be stopped Histogram publishes The histogram objects in Image publish the following messages 8 8 Publish and Subscribe Image 2 1 Library User manual SPB_DESTROY is published when the object is about to be destroyed The data with this message is always a NULL pointer SPB_CHANGED is published when the histogram data changed The data with this message is a NULL pointer Error stack publishes In the Image library an error stack is present to log the location and messages of an error occurring somewhere in the low level of the image processing functions When an error occurs the relevant functions handle it and all add their error values and messages to this stack When all these functions have aborted the error stack publishes a SPB_CHANGED message to signal to the interface that an error has taken place The stack is a global structure called im_error_stack a pointer to it can be retrieved using the function get_im_error_stack The type of the structure is IM_ERROR_STATUS that is defined in the include file im_error h Please note that this variable is a structure and in order to subscribe to it you must use the address of this structure The correct subscribe call for the error stack therefore is include spublish h include im_error h IM_ERROR
240. ootnote Text Badd one Footnote Text Kadd one MY FUNCTIONS Footnote Text USER_FUNCTIONS 000029 _ Page 1 Sec1 1 1 At Ln Col 1 REG MEK EXT OVA WEH Save it and start the Help Workshop from the Microsoft Visual C environment In the Help Workshop open the file scilimag hpj from the help directory To compile the help file Select File Compile Compilation will take some time the help compiler will minimize itself during compilation and restore itself to normal view after compilation It generates the new scilimag hlp which makes your Help page reachable from SCIL_Image like that of any other command Getting started 3 35 SCIL_Image 1 4 User Manual amp SCIL_Image 1 4 Windows Manual Iof x File Edit Bookmark Options Help cones rder on Bie lt gt NAME add_one add 1 to an image SYNOPSIS include user _def h Hinclude image h int add_one IMAGE in IMAGE out DESCRIPTION Add 1 to the image in and store the result int the image out RETURN VALUES OK 1 if successful NOT_OK 0 ifnot successful AUTHOR written by M Y Self 19970204 Figure 3 17 Your new manual page We have used the example to show you that it is possible to embed your own function as a command in SCIL_Image rather than as an exact explanation of all steps involved We will give that in Chapter 5 You should remember these salient points e You can embe
241. op would result in a size change for a ROI a warning will be issued and the operation will be aborted A call to pre_op resulting in a type change for a ROI is no problem If a ROI is specified as both the input and output of an operation pre_op will copy the data from the parent to the ROI image as described above and change the type of the ROI if needed Multiple calls to pre_op As explained above after a call to pre_op an IMAGE structure can contain two different images This is the case until a call to post_op is made with the image Before that time other calls to pre_op using ADJUST in which the same image is specified as being an output image will fail This can happen if in the routine a call to another image processing routine is made after the pre_op call because the other routine is also likely to call pre_op This must be prevented Calls to other image processing routines must be made BEFORE the pre_op function or AFTER the post_op function 9 18 Programming with Image Image 2 1 Library User manual The post_op function A call to post_op is issued at the end of each operation for each image that was specified as an output image in a call to pre_op using ADJUST _NIP post_op requires one parameter int post_op IMAGE image post_op cleans up after the actual calculations of a routine have been done The function checks if the input a
242. or SCIL_Image You can find help from the Contents page or by using the Search menu An example of a Elp page is that of copy_im amp SCIL_Image 1 4 Windows Manual OF x File Edit Bookmark Options Help cones inden o Be e gt NAME copy_im copy image SYNOPSIS include im proto h int copy_im IMAGE in IMAGE out DESCRIPTION Copy image in to image out NOTE For more powerful image arithmetic expressions scaling adding offsets etc use the function eval RETURN VALUES IE _ OK 1 on success Negative error status on failure see im_error h SEE ALSO copy part image The C Interpreter 4 13 SCIL_Image 1 4 User Manual The information is in a format similar to that of a C function on a UNIX system which is rather self explanatory even if you have not seen it before The green underlined terms provide links to other related help pages Sometimes you can obtain a popup explanation of terms that are dashed underlined You can also add help pages about your own functions and add them to the standard Help facility You do this by the procedure given on the Help page User defined commands They are then reachable in the same way as the functions we put in Ctrl Enter on selection From any text window you can pop up a dialog box of a command by selecting the command using the mouse double click on it pressing Ctrl Enter Once on the dialog box selecting
243. or on these streams It is up to the user interface to present the text in a suitable layout Function overloading Image uses the mechanism of function overloading for flexibility maintainability and simplicity The mechanism works by having the top level image processing functions look at the type of the image and calling a low level functions that is tailored for that type of image Here we consider how that mechanism works and is implemented in Image With use of the function overloading mechanism new image processing routines can be implemented for several types of images or existing operations can be implemented for additional image types Three layers In order to make use of the function overloading the source code should be divided into three layers 1 generic function level 2 parameter checking and type and size adjustment 3 processing of the data By splitting the processing of the image in layers 2 and 3 the actual processing function layer 3 can be used for more than one image type If the image adjustments and parameter checking layer 2 differ per image type this is especially useful Consider the function add_im as an example to describe the three layers add_im adds the values at corresponding locations in two images and stores the result in another image The call to add_imis add_im A B C A and B are the input images and the result is stored in C The images A and B are GREY_2D in this exampl
244. ord compared with the other image types The t ype_ident must be a unique number that identifies the image type In the header file 1mage h the type_specs and type_idents of the standard image types are listed For the byte 2d image type we choose at ype_spec of 1024 10th bit set and a type_ident of 11 Both of these values are currently unused in the standard set of image types A structure with image type specific information must be created This structure will be linked to the in_descript and out_descript pointers of the IMAGE structure The first two fields of this structure must be an int that contains the type_ident of the image type and a pointer to the image data BYTE The dimensions of the image must also be stored in this structure The byte image type does not need any additional information so the structure is complete It is advisable to define macros to access the individual fields of the structure which can be used in the source code The defines the structure and the macros are best put in a file which is named after the image type BYTE is already defined in image h In this example we store them in the file pyte_2d h ifndef BYTE_2D_H necessary to avoid multiple definitions define BYTE_2D_H file inclusion define BYTE_2D Ti unique value define BYTE_2D_SPEC 1024L unique bit field New image types 12 3 Image 2 1 Library User manual imag
245. orrect the error and continue or issue a warning and discontinue processing The value indicating success in Image is E_OK which is defined as 1 Since various things may go wrong a failure status may be indicated by one of many values so that the programmer is able to indicate the cause of the error In Image the failure status E_NOT_OK value 0 simply indicates that an error occurred but gives no information as to the cause A negative value is used to indicate the cause of the error Those used in Image can be found in the include file im_error h In some cases the negative error status can be ambiguous if for instance a function returns some kind of measurement value this value may well be negative even when no error has occurred Other functions may be defined as being unsigned always positive or returning a pointer In these cases the caller can use the function im_get_status to inquire if the called function exited with an error or with an OK status im_get_status will always return the error value of the function that was most recently called by the current function provided that the function participates in the error mechanism When no error has been reported I E_OK is returned Error handling When detecting an error it is common practice to abort the normal processing of an operation and report the error to the user Because of the independence of any user interface functions in Im
246. ould be clear that the offset is mirrored in the border of the image If however border value is specified to be any other number say 5 then no mirroring will take place but all pixels addressed outside the image will have the specified value To illustrate eval C A 256 0 5 Aside from the use of operators listed in Table 6 3 eval can be used to evaluate a number of mathematical functions in the expression For instance eval C sqrt A calculates the square root of every pixel in image A and stores the result in image C Functions can be mixed freely with operators eval D sqrt A 255 The result of the operation sqrt is a float so the resulting image will be a FLOAT_2D image If however a GREY_2D is wanted the data can be converted by using the function irint in the expression eval D irint sqrt a 255 6 20 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual The following functions may be used with eval function description _type of result abs x absolute value of x float acos x arc cosine of x float asin x arc sine of x float atan x arc tangent of x float atan2 x y arc tangent of x y float cbrt x cubic root of x float cos x cosine of x float exp x to the power of x float hypot x y square root of x x y y float irint x returns x as a integer integer log x natural logarithm of y float log10 x base logarithm of y float log2 x base logarithm of y float max
247. ow about the menu system and the command level Selecting an item from a pull down menu will result in interaction with the system via a dialog box and subsequent execution of the command Reading Chapters 1 through 3 should be sufficient for this level of use although Chapter 4 may also be useful On the other hand if the system is used for the development of new routines and applications it will be necessary to become familiar with the system s philosophy and C In fact the step from a sequence of commands typed in the window to a complete C program is small Advanced use of C and the image data structures may boost development greatly When developing algorithms and applications reading Chapters 4 and beyond is recommended Summary of the Chapters Part I of the manual provides a general orientation Chapter 1 gives a global description of the SCIL_Image system Chapter 2 explains how users should set up their personal environment and obtain a personal version of SCIL_Image Chapter 3 describes how to start using the system and presents sample sessions which allow users to get some experience using the system The chapter ends with a discussion on some general design issues such as the look and feel of the user interface Part II is about the interaction system SCIL and Image as part of SCIL_Image Chapter 4 describes the C interpreter SCIL It explains the direct command mode the program mode the history mechanism the command
248. playing that image If no table is attached explicitly to an image a normal grey scale for grey float and complex images is used Binary and labeled images do not need a lookup table because the method for displaying them cannot be modified The available lookup tables at start up are The Image 2 1 library in SCIL_Image 6 7 SCIL_Image 1 4 User Manual name contents EMPTY_LUT all entries black BLUE_LUT a blue scale GREEN_LUT a green scale CYAN_LUT green and blue scale mixed RED_LUT a red scale MAGENTA_LUT red and blue scale mixed YELLOW_LUT red and green scale mixed GREY_LUT normal grey scale LABEL_LUT scale that simulates label display MULTI_LUT primary colors each in a specified bit OVERLAY_LUT_1 8 FALSE _COLOR_LUT specified bitplane red false color table Table 6 2 default color lookup tables A lookup table can be attached to an image as follows set_clut A OVERLAY_LUT_8 The lookup table OVERLAY_LUT_8 is filled with a grey scale with lookup values ranging from 0 255 For each value for which the binary representation has the 8th bit set the corresponding color is red This results in all values that are over 128 being displayed as red To get the grey scale back again set_clut AGREY_ LUT Image management Aside from manipulating the display of an image it is sometimes necessary to create a new image or change the image type For instance to create a test image on which to try out a certain
249. processing done is limited an image is read thresholded labeled and some features of all the objects in the image are measured and printed 1 The pointer to the top level image is retrieved this object publishes the creation and destruction of images as discussed in Publish and Subscribe in the Image library on page 8 5 2 The pointer to the error stack is retrieved This stack can be used to get detailed information on an error occurring in the library More information on the error stack can be found in Error handling and reporting on page 9 21 Creating an Image Application 13 3 Image 2 1 Library User manual To redirect text output a function can be registered Details of which can be found in Textual output on page 9 26 The function itself analyses the st ream parameter and if it is a special stream it prints a header above the text indicating which stream The im_handle_error function is subscribed to the error stack When an error occurs and the library returns this function is automatically called and prints a stack trace of the location of the error The handle_super_im function is subscribed to the top level image When an image is created this function prints a message that a new image is created and then subscribes the handle_images to the new image handle_images in its turn will print out information about the publishes that come from these imag
250. pter Image 2 1 Library User manual What is Image2 1 Image 2 1 is a portable image processing software library written in the programming language ANSI C It is currently available as a stand alone C library on a variety of computer platforms UNIX PC MS Windows Macintosh Key features of Image2 1 are e a flexible image infrastructure e support of a large set of basic image types e an advanced and extensive set of image processing functions e fast implementations e abstract error and I O handling e publish and subscribe mechanism The image2 1 library is not a complete image processing development environment as it lacks a real user interface and viewing capabilities The library was developed to provide a framework for developing image processing software The library does not contain system specific routines and as such it is easily portable to other platforms Image infrastructure The image library provides a framework to support a variety of image types The infrastructure maintains a strict separation of image operations by type The infrastructure provides a convenient and image type independent way to do image housekeeping It deals with input output adjustment of image operations and manages region of interest definition in images The type mechanism simplifies the management of images and makes programming with different types of images easy Image types Current supported image types are e Grey
251. r condition it calls im_end_func also with its name as the parameter When no error occurs the name stack grows and shrinks as functions are being executed and returned Programming with Image 9 21 Image 2 1 Library User manual If a function detects an error situation this is reported to im_report_error and an error status is returned to its caller It specifies its name an error value and an optional information string as the parameters Once an error is reported the name stack is copied to the error stack and at that moment the exact location of the error is known While all the functions in the stack handle the error and clean up they store additional information in the error stack Other functions than the ones in the error stack trace are not allowed to store information in the error stack so only information regarding one error is stored When all functions have returned the error stack publishes that an error has occurred and the user interface can inform the user about the error if necessary Return values of functions Functions written for use in Image should return a value based on their success or failure The value can be examined when the function returns and its caller can take various actions depending upon the return value Usually if the function returns an OK status the processing should simply continue If a problem occurred the status should indicate which problem arose The program can either c
252. r implied are hereby disclaimed specifically including but not limited to express or implied warranties of merchantability or fitness for a particular purpose Chapter 1 Outlines of SCIL_Image wu cccsssssssccescsssssscccsssssssecsees 1 1 Whe four layers of SCM Mae a Gaaen cheats teden sane E EA R mona etna 1 2 Using the system at the application level eee eeeseceseeeeeeeeeceeeeeeseeeeeeteeeesaes 1 2 Using the system as an interactive User o c0ict ns cg thincvsas aces dncudeaeeaaensdncpetsoeeaetens 1 2 Using shorthand typing for application development eee eeeeeeeeeeereeeneeees 1 3 Contr l statementS zana faaas cand get baaerapaton gate a sahara chub daa Saatele euse 1 3 Programming an image processing fUNCTION eeeeeeeeeeeeceeeeeceeeeeceteeecneeeesaes 1 4 Making your own brary i650 sir Ba Gee eg eta Bleed ees 1 5 Buildi g pplicati ns s si ae ce opdeneteheds saa aa i saia is 1 5 SCIL and the Image Brar iiaa a a S R K 1 5 O teadis this majua occ sco E A Ee a a eea E E A EaR en iaa 1 6 Summary ofthe Chapters eo i EE TE R EER 1 6 Wy Ey SCIE Image aes oes a a E ute sa S E A A A 1 8 A multi level interactive processing environment sesesrsesesressrseresrersersrerressesee 1 10 SC Lia ry Mama Cl sede ceshces A vcnsave a waco ta su ceo nce eae E RS 1 10 o s DBE Gt TS ge cele eaten center ee men nn a te mre REN Ee a mn R s 1 11 SC IE Command expander oscoacicsstsaasssesssecuessseceanieauaaceaisencnesseees a
253. r of the trui image Type in the commands readf trui A thresh A B contour B Open a new file with File New C Program This pops up a RunEdit window Copy the last two commands from the Worksheet into the RunEdit window using Edit Copy Save the file as contourA mac in the directory scilimage14 demo macro You can execute the commands in this macro by typing macro contourA mac The macro makes the contour of any grey valued image in window A and displays the result in window B In this way you can collect useful command sequences However you cannot give parameters to macros so if you would need a macro to make the contour of the image in B you would have to write another macro In that case it is better to write a program see the next section Session Four Programming in SCIL_Image Macros are very useful for writing demos Getting started 3 25 SCIL_Image 1 4 User Manual which use the same images and display results in the same way or for saving effective sequences of commands from a session This concludes the second sample session To quit SCIL_Image type exit Session Four Programming in SCIL_Image In this fourth session we will build a small program We show you how to incorporate this program as an interpreted command into SCIL_Image We also demonstrate how to bring this program into the SCIL_Image menu system The program will read an image from a file thresho
254. ram The include file image h contains the definitions of the MAGE and PIXEL data types and also of the pre processor macros ImageInData ImageOutData and ImageSize The use of these macros is explained in the chapter Programming with Image for now know that the first two return the pointers to the input and output images and the third returns the size of an image 3 30 Getting started SCIL_Image 1 4 User Manual In itself this function can be used but it is not yet properly embedded within the SCIL_Image structure If you execute it you will need to call the function display_image explicitly to display the image To embed the function properly into SCIL_Image we add some lines add_one c version 2 pre_op post_op added include image h include im_infra h g_add_one IMAGE in IMAGE out PIXEL indata outdata long num if pre_op in out ADJUST G_2D_SPEC GREY_2D return NOT_OK indata ImageInData in outdata ImageOutData out num ImageSize in while num outdatat indatat 1 return post_op out The functions pre_op and post_op do the administration involved in making sure that your function g_add_one works on images of type g2D They will pop up error messages if you try to use the function in ways you did not specify If your operation could no
255. ration from A to B Before executing the filter change the filter size from 5 to 9 to make the effect more pronounced choose Filter Size 9 click OK You will have to wait a little while since this is quite a complex operation During the operation the status indicator in the lower left corner changes to Running and an hourglass icon appears next to it Getting started 3 13 SCIL_Image 1 4 User Manual This may be a good time to experiment with some of the filters in Image Filter to acquaint yourself with the standard capabilities of SCIL_Image If you want a description of a command just press Help in its dialog box and a Help page will pop up Forkuwahara this reads amp SCIL_Image 1 4 Windows Manual OF Xx File Edit Bookmark Options Help covers index E em lt gt kuwahara NAME kuwahara edge preserving smoothing Kuwahara SYNOPSIS include im proto h int kuwahara IMAGE in IMAGE out int fsize DESCRIPTION Perform an edge preserving smoothing Kuwahara filter on the pixels of mage n and store the result in image out Image in is scanned with a moving window with dimensions fsize fsize The command subdivides the moving window into four sub windows In each sub window the variance of the pixel values is calculated The window with the lowest variance is taken as the averaging window The resulting average value is stored
256. ration with other applications It becomes increasingly popular to include the image processing software as a component into more complex applications or in other words as plug ins into other applications Furthermore 7 4 Introduction to Image 2 1 Image 2 1 Library User manual creating end user applications with its own user interface and look and feel is essential for many companies and institutes Therefore there is a clear need for a stand alone image processing library The library provides a framework which makes it easy to develop image processing software As the library contains no system specific calls and maintains a strict abstraction in the user interface the library is platform independent and suitable to integrate into virtually every application Structure of the Image library In the Image library it was decided to maintain a strict separation of image processing functions by the type of image they work on Associated with each image type are a number of type dependent operations such as image arithmetic and image I O operations The default supported image types are image type name in Image grey valued 2 d GREY_2D grey valued 3 d GREY_3D binary bitmapped 2 d BINARY_2D binary bitmapped 3 d BINARY_3D floating point 2 d FLOAT_2D floating point 3 d FLOAT _3D complex 2 d COMPLEX _2D complex 3 d COMPLEX _3D color 2 d COLOR_2D color 3 d COLOR_3D labeled 2 d LABEL _ 2D labeled 3 d LABEL_3D Table 7
257. rations For the BITMAP algorithms only the results using the logarithmic decomposition are shown Discussion and Conclusions In this chapter we presented new algorithms for binary morphological transforms based on a bitmapped representation of binary images and the use of logarithmic decomposition of 11 10 Bitmapped binary images Image 2 1 Library User manual structuring elements The new algorithms prove to be on average one order of magnitude faster than existing fast implementations The results presented in this chapter indicate that bitmapped representation of binary images is well suited for morphological image processing as it is very efficient both in terms of memory requirements and in terms of algorithmic efficiency Instead of the necessity to allocate at least 8 binary images when using the pixel mapped representation only the bits in one binary image are stored When working with very large images like those acquired with page scanners the reduced memory requirement is an important feature In order to deal with large structuring elements obtained from auto dilation of smaller structuring elements logarithmic decomposition proves to be a powerful tool Literature HARALICK R M Haralick S R Sternberg X Zhuang Image Analysis using Mathematical Morphology IEEE Transactions on Pattern Analysis and Machine Intelligence PAMI 9 532 550 July 1987 MARAGOS P Maragos Tutorial on Advances in
258. re available to detect the occurrence of an error 1 Checking the return value of the function it called from the Image library This will be most common because the program can react to the error situation immediately when it occurs an that may be necessary to control the correct execution of the program 2 Subscribing to the error stack is typically done by the user interface because it enables a uniform mechanism from a central location to visualize the occurrence of an error to the user Checking routines Image provides a number of routines to check the values of non IMAGE parameters Each of these functions checks whether a particular requirement is satisfied and if not reports this as an error to the error stack If the requirement is met IE_OK 1 is returned otherwise IE_NOT_OK 0 is returned These functions can be used in an i statement to allow the programmer to take appropriate action or prevent further processing if the routine fails The number of check routines is too large to discuss each of them here An example of how they are used is shown below followed by a list of the routines provided For a full description of these functions the reader is referred to the on line reference manual As an example we consider a function my_filter with the following header Programming with Image 9 23 Image 2 1 Library User manual int my_filter IMAGE in IMAGE out int size int of
259. rences between these terms We call the last two possibilities the Direct Command Mode because you type them into the Worksheet The C Interpreter 4 3 SCIL_Image 1 4 User Manual Commands in the Direct Command Mode are very much like lines in a C program both in syntax and in meaning You can combine several commands on one line using the same rules as in C More specifically the syntactic possibilities for commands given in the Direct Command Mode are e Classes of commands e Variable declarations Ine Gs e Statements for i 0 1 lt 3 1i printf sd n i e File inclusion directives include image h e Pre processor defines define max a b a gt b a b e Type definitions typedef struct date_t int month year DATE DATE date date month 2 date year 1993 e C functions In SCIL_Image functions are invoked in the same manner as in C The functions used throughout this section are part of the standard C library You cannot define your own functions in the Direct Command Mode you would always have to define those using an editor because a program is a named collection of commands There are ways to do this in the other modes Once you have defined your own functions you can use them in the Direct Command Mode if they are e macros see The Macro Mode on page 4 5 e loaded interpreted functions or UFOs see The Programming Mode Interpreted C functions and UFOs on page 4 7 e compil
260. reshold Edit When you are satisfied you may exit by clicking OK This executes the thresholding in image A with the settings chosen a binary image is the result If you cannot find a good threshold and therefore would not like to do the thresholding operation you should exit the tool by clicking Cancel which returns A to its original grey values The final interactive tool is the Grey Map Editor select image A click the Grey Map Editor icon mE E The Grey Map Editor interaction window displays a function which maps grey values of image A to other grey values You can change this function by grabbing the nodes of the graph and dragging them to add or delete nodes click the right mouse button All displayed copies of image A change accordingly When exiting the Grey Map Editor you have the choice to keep the changes to the image OK or to restore the original grey map Cancel Note through all of these actions that SCIL_Image allows you to have multiple interaction windows open and does not require you to close one before you can interact with the next Its way of interaction is modeless This concludes the first sample session Should you want to stop now select File Exit SCIL If you still had the histogram of A displayed you will notice that it becomes void and displays the error message IMAGE TYPE DOES NOT SUPPORT HISTOGRAMS 3 12 Getting started SCIL_Image 1 4 User Manual Session Tw
261. rom menus or in the Direct Command Mode they have thus become actual SCIL_Imagecommands Also you can use them as building blocks for more advanced functions There are three files which describe new functions or programs to SCIL_Image apart from your own files that define what they actually do They are scilimage14 comfile scilimage14 prog sysfunc c and scilimage14 prog overload c You may not need to generate all three files depending on what kind of functions you want to add e comfile If you want to extend SCIL_Image with your owninterpreted programs all you need to do is generate a new scilimage14 comfile This file will contain information on the menus functions default values and dialog boxes for all commands in the standard SCIL_Image version and for your interpreted additions You supply this information through a Command Description File CDF for short recognizable by the suffix cdf In Session Three Using the SCIL_Image Command Line Mode page 3 21 this file was called personal cdf We give details of the CDF file in Making a Command Description File CDF on page 5 3 e sysfunc c If you want to make a new version of SCIL_Image that contains your owncompiled functions as new commands in SCIL_Image you also have to make a file scilimage14 prog sysfunc c which contains references to all functions including your own This file is generated by mksysfnc started by the nmake
262. rt if xtab amp amp ytab fatal_err l_byte_2d_disp_8bit fill the size stretch tables 12 10 New image types Image 2 1 Library User manual skip_tab xtab xsize dispw skip_tab ytab ysize disph f i get the destination position of the pixels to be displayed part_sx double sx xsize dispw part_sy double sy ysize disph part_width double width xsize dispw 0 5 part_height double height ysize disph 0 5 point to the first destination position swap swap part_sy dispw part_sx for i 0 i lt part_height i pix BYTE Byte2dImageData ip xsize ytab part_sy i vdi swap swap dispw xt amp xtab part_sx j part_width while j gt 0 tmp pix xt vditt BYTE dtab tmp free xtab free ytab PwWrite vport part_sx part_sy part_width part_height im_end_func 1_byte_2d_disp_8bit return IE_OK Image type information In the SCIL_Image environment the information supplied to the user in the title bar of an image window and the information in the floating window when the left mouse button is pressed in an image auto_point information is supplied by two small functions In the title bar the name of the image its type and the sizes of the image are generally displayed The auto point information supplies the user with coordinates an
263. rt function int byte_2d_copy_masked_part IMAGE in BOOL_MASK mask IMAGE out int sx int sy int sz int width int height int depth int dx int dy int dz int clear T BYTE ptr int status im_begin_func byte_2d_copy_masked_part if pre_op in in COMPARE BYTE_2D_SPEC ImageTypelIdent in pre_op out OUT ADJUST ImageTypeSpec out ImageTypelIdent in return im_report_error byte_2d_copy_masked_part IE_PRE_OP if ptr l_byte_read_masked_part ImageInData in mask SX Sy Sz width height 1 ImageWidth in ImageHeight in ImageDepth in return im_report_error byte_2d_copy_masked_part IE_NOMEM Unable to allocate part image buffer 1_byte_write_masked_part ImageOutData out ptr mask dx dy dz width height 1 ImageWidth out ImageHeight out ImageDepth in clear free ptr if status post_op out lt IE_OK return im_report_error byte_2d_copy_masked_part status im_end_func byte_2d_copy_masked_part return IE_OK Displaying the image To view the image on the monitor in the SCIL_Image package a function must be implemented to display the image The function must meet the following requirements e Jt must be able to display an arbitrary rectangular part of the image needed for ROI display The parameters of the function specify the image the orig
264. s Floating point images consist of pixels each of which can have any real value within precision limitations Each pixel is represented by the C type float which on most computers occupies 32 bits Though floating point images can contain more information than integer valued images they are far less common This is because computers used to work far more efficiently with integers and because integer valued images require half the space 16 bits per pixel and most image acquisition happens in integer valued format 9 4 Programming with Image Image 2 1 Library User manual Complex images Each pixel in a complex image is a complex number and therefore has a real and imaginary part In Image each of the parts is represented with the C type float This type of image is specifically required for the Fast Fourier Transforms FFT Labeled images Like a grey valued image a labeled image is made up of integers with the C defined datatype PIXEL The interpretation of the values in a labeled image is however entirely different The information in a labeled image is similar to that in a binary image The pixels in the image either belong to an object or to the background In a labeled image however each object in the image has its own number or label used to distinguish it from other objects in the image The pixels in an image which have been determined to be part of an object and are connected are all as
265. s see Table 9 1 present image types GREY_3D and LABEL_3D The function pre_op itself returns OK 1 if the images are of the same class and have the same sizes The fault status of pre_op is NOT_OK 0 Because pre_op handles ROI processing all operations should call pre_op because the specified image can always be a ROI For example the routine that writes an image to disk writefile has only one image as input and can handle all types The call to pre_op therefore is pre_op image image COMPARE WHATEVER OUT_AS_IN but could also be pre_op image image COMPARE ImageTypeSpec image ImageTypeldent image Both calls have the effect that all image types are accepted by pre_op If the image is actually a ROI it will be written to disk as if it were a normal image In the case that an operation has two input images which may be of different types but must be of the same sizes the call to pre_op would be pre_op inl in2 COMPARE ImageTypeSpec inl ImageTypeldent in2 ADJUST _NIP mode The ADJUST _NIP mode of pre_op is more complex than the COMPARE mode because its behavior can be influenced by some global variables If pre_op is called using ADJUST the output image will be split up if the output image has to change type or size as is explained above When the output image is of the correct type and siz
266. s is a value is in the specified range frange_ok checks is a float value is in the specified range odd_ok checks if a value is a odd integer value even_ok checks if a value is a even integer value bit_ok checks if a value is a binary value different_ok checks if two values differ from each other positive_ok checks if an integer value is positive fpositive_ok checks if a float value is positive greater0_ok checks if a integer value is bigger than zero fgreater0_ok checks if a float value is bigger than zero unequal0_ok checks to see if an integer value is unequal to zero funequal0_ok checks to see if a float value is unequal to zero power_of_2_ok checks if a value is a power of 2 Please refer to the on line manual pages for a complete description of these functions Apart from these checking functions that generate an error also some testing functions exist that only return a true false value and not generate an error These functions all start with the is_ prefix they are is_image checks if the pointer supplied is a valid image pointer is_clut checks if the pointer supplied is a valid color lookup table Check_image_integrity The function check_image_integrity checks all images for irregularities For instance if the input and output descriptor of an image are different it will delete the image connected to the output descriptor and reconnect out_descript to the same descriptor as in_descript see also Figure 9 1 and F
267. s listed above can be found in the reference pages for these functions Programming with Image 9 43 Chapter 10 Analysis of Images and Objects AIO This chapter describes the library for Analysis of Images and Objects AIO AIO is an infrastructure for interactive and automatic object measurements It provides mechanisms for object selection and for storage of objects in image silos The general concept is discussed and examples of the use of AIO are presented Do not read this chapter e If you are a beginning user e If you have never programmed in C You should read this chapter e If you are an experienced user with a basic knowledge of C programming e If you want to perform object measurements Image 2 1 Library User manual Introduction AIO is a framework for the analysis of objects in images It is especially suitable for performing measurements on microscopical objects Objects are manipulated either on the basis of the measurements or by pointing at their images The object data vector and the object image can be stored in a file and in an image silo respectively for later retrieval The combination of AIO and Image is a flexible environment for image analysis and application development General Concepts in Microscopical Image Analysis Most microscopical applications measure certain features of objects in images and or classify objects into different categories and therefore adhere to
268. s to the object An object structure should look like this typedef struct my_object void publish publish field other fields 8 2 Publish and Subscribe Image 2 1 Library User manual The name for this field publish is not mandatory just recommended Subscribing and unsubscribing to objects To receive the messages broadcasted by a publishing object the receiving object must subscribe to that object For that it must use the function spb_subscribe void spb_subscribe void object void ident void fun_ptr void client_data object is the publishing object ident is an identifier for the subscriber typically a pointer to the subscriber fun_ptr is a pointer to the function the subscriber uses to receive the messages client_data is an optional pointer to data the subscriber wishes to receive whenever the publisher broadcasts an event For a complete description of the function fun_ptr see Receiving messages on page 8 3 After subscribing to an object the subscriber will receive all messages send by the publisher and may or may not react on each of the messages After a while the subscriber can decide that it is no longer interested in the object and wishes to unsubscribe For that it calls the function spb_unsubscribe void spb_unsubscribe void object void ident void fun_ptr object is the publishing object it is subscribed to ident is an identifier for the subs
269. sages of super_im SPB_ITEM_ADD is published when a new image has been created the data of this message is a pointer IMAGE to the newly created image SPB_ITEM DELETE is published when an image is about to be destroyed The data is apointer IMAGE to the image When this message is published nothing can be done to stop the destruction of the image The only reason this message is published beforehand is to prevent invalid memory access by subscribers that still reference the image because they do not yet know that the image no longer exists SPB_NEWTYPE is published when the type of image changes The data is a pointer IMAGE to the image that has changed Regular images publish this message during the call to the post_op function However Region Of Interest ROD images publish this message during the call to the pre_op function SPB_RESIZED is published when the sizes of image changes The data is a pointer IMAGE to the image that has changed SPB_AUTO_DISPLAY is published when using the functions don doff auto_display The data is a pointer int toa Boolean value that states if the auto display mode is on 1 or off 0 The auto display mode is considered to be a hint to the GUI to not display the images when they change This feature is used to hide intermediate results of complex image processing operations Image publishes After obtaining the pointer to an image an obj
270. se it to edit text in any text window or to move text between windows Getting started 3 37 SCIL_Image 1 4 User Manual SCIL chain macro load tun list more logon logoff myar dyn_link dyn_unlink Figure 3 20 the SCIL menu The SCIL menu contains commands necessary to the functioning of SCIL_Image as a programming environment These commands are described in detail in The C Interpreter chapter 4 chain macro load run are commands to manipulate interpreted programs in SCIL_Image list logon logoff rmvar are commands used for program development to show and record interpreted programs in SCIL_Image Image 10 Histogram Generation Manipulation Filter Transform Segmentation Conversion Single_Objects Statistics Var_Objects Morphology Texture a a Oe eS ee Se a es a a a convert roi_define show_image_info show_func_overload copy_part_image destroy_image eval copy im make_image all_im Figure 3 21 the Image menu 3 38 Getting started SCIL_Image 1 4 User Manual This menu contains the image processing commands and is therefore the menu you will probably use most It contains many sub menus I_O contains functions which read and write images from and to your disk Histogram has commands to manipulate image histograms Generation has commands to generate test images with simple geometric and grey val
271. selected from the menu they can only be typed in the text window The SCIL_Image system uses the syntax of the C programming language in specifying the control commands A control command is typed in the text window in the standard C syntax Upon completion by the lt return gt or lt enter gt key the statement is interpreted and executed The C interpreter will check for syntax availability of the requested function and then issue the requested function For an example see the guided tour in Getting started on page 3 3 1 Outlines of SCIL_Image 1 3 SCIL_Image 1 4 User Manual Conceptual levels Modes of operation menu bars Interactive operation with menus and SCIL menu system Command Description File dialogue boxes text window SCIL command expander Command Expander and C interpeter SCIL C Interpreter SCIL library handler K D User defined or SS imported libraries Image libraries Figure 1 1 The layers of SCIL_Image and its modes of operation Programming an image processing function Going one step further in expertise the need may arise to developed a new image processing function This can be done by programming at the level of the C interpreter Newly developed functions can be made available quickly in the SCIL_Image system by inserting the necessary function identification information in the Command Description File Restarting the system is sufficient to integrat
272. session introduces the various aspects of command line operations in SCIL_Image It teaches you how to give commands by typing rather than by using the menu You will need this ability to write programs in SCIL_Image Please work through this session by keying in the examples We assume that you have completed the previous sample sessions Restart SCIL_Image if it is not still active double click the SCIL_Image icon In the previous session we showed how SCIL_Image can be operated using the menu system In this session we will use SCIL_Image as a C interpreter C being the programming language This means that if you type statements with correct C syntax the system will execute them without the usual sequence of compiling and loading normally required when programming in C For instance to have the system print some text you can type this C statement copy this line precisely including the semicolon in the window Worksheet printf Hello world n hit the Enter key on the lower right of the numeric keypad The printf line is a legal C statement which will be executed immediately after you hit the Enter key The result is that the Worksheet shows Hello world We will use bold italics to indicate program output Notice the difference between the use of the two Enter keys on your keyboard in all text windows of SCIL_Image Enter on the keyboard the Return key gives a new line does not e
273. set_var_object_type sets the type of a var_object to a specific type set_var_object_size sets the dimensions of a var_object to specific values set_var_object_data sets both the type and the sizes of a var_object set_var_object_class sets the class of a var_object set_var_object_comment attaches comment to a var_object Several useful functions are available for inspecting and manipulating the var_object as a whole list_var_objects lists all var_objects that exist show_var_object_info shows information on a var_object dump_var_object dumps the contents to the terminal or a file copy_var_object copies a var_object to another one write_var_object saves a var_object to disk read_var_object reads a var_object from disk These functions have a on line manual page that contains information on them Checks on var_objects The function is_var_object can be used to determine the existence of a var_object It returns the value l if the var_object exists and 0 otherwise The function var_object_ok does the same and pops up an alert box if the var_object does not exist Conversion of var_objects to images and vice versa Var_objects are implemented to simplify the management of non image data for the user As a var_object may contain data that could also be stored in an image it might be useful to process the data in a var_object with one of the image processing functions To make this Programming with Image 9 39 Image 2
274. signed the same unique label Each pixel in a labeled image has a value identifying it as a member of a set or object Labeled images allow measurements such as size and shape to be performed on the image objects Color Images In Image color images can be represented by five different color models The present models are RGB CIE XYZ CIE Lab HSI and CMYK The color models are each represented in a different manner RGB each pixel is made up of a Red Green and Blue component each component channel occupies 1 byte in memory C type unsigned char The three channels together with a fourth extra and empty byte are stored in one C structure CIE XYZ each pixel consist of three floating point values X Y and Z that are stored in one C structure Each channel occupies 4 bytes of memory C type float CIE Lab each pixel consist of three floating point values Lightness a and b that are stored in one C structure Each channel occupies 4 bytes of memory C type float HSI each pixel consist of three floating point values Hue Saturation and Intensity that are stored in one C structure Each channel occupies 4 bytes of memory C type float CMYK each pixel consist of four floating point values Cyan Magenta Yellow and K sometime also named Blackness that are stored in one C structure Each channel occupies 4 bytes of memory C type float Programming with Image 9 5 Image 2 1 Library User manua
275. sion of an image in bitmap A resulting in an image in bitmap B The structuring element is encoded in the arrays X Y and F EROSION A B X Y F DO i LineOffset index the first word in the actual image FOR y 0 TO L 1 DO Lis the size of the image in y direction FOR x 0 TO n 32 DO result BITWORDALL all bits set to 1 FOR j 0 TO Noels 1 DO IF F j gt 0 result result amp BARRELSHIFT A i X j Y j 1 ENDDO ENDDO B i result i i l index the next word ENDDO i it l skip over the words in the border ENDDO Evaluation In this section the performance of the proposed algorithms will be compared with the performance of existing implementations The new algorithms will be referred to as the BITMAP algorithms They will be compared with two software implementations and with a Special purpose Image Processing machine KONTRON which will be referred to as SIP All our BITMAP algorithms are coded in portable C Tests were performed using a SUN SPARC station 1 except for the results obtained with the SIP All the timing results are obtained by averaging the results over at least 20 experiments Because some implementations of the morphological transforms are data dependent we have chosen a set of three test images These are depicted in Figure 11 4 11 8 Bitmapped binary images Image 2 1 Library User manual fo i ai ae TE E g a j mi ee J he Figure 11 4 Test Images a Trui 256x
276. specified in the minimum field Only var_objects that have that name in the object_class string will appear in the dialogue box e comment is a pointer to a string of arbitrary length This string can be used to store a comment concerning the var_object When the var_object is written to disk the comment is saved in the header file of the var_object The macros should be self explanatory but a note on the V_O_Length macro may be in place It is used to determine the size of the entire var_object in bytes To be specific it is the product of the fields nr_channel size and bpelem 9 38 Programming with Image Image 2 1 Library User manual Programming with var_objects The var_objects are designed to be an easy interface to C arrays for the user Any operation that deals with arrays of data that must be accessible to the user should use var_objects The var_objects created are stored in a linked list When programming a function the var_objects can be used like normal arrays once they have been created Low level functions however that need arrays may also use var_objects rather than arrays The advantage is that their size and type can be easily adjusted To dynamically adjust the type and the sizes of var_object several functions are available var_object creates a var_object and returns the pointer destroy_var_object destroys a var_object var_object_by_name retrieves the pointer to a var_object through its name
277. ssing has been done the grey image is no longer valid and is therefore removed by post_op resulting in a conversion of the image type post_op frees the memory of the grey image throws away the input descriptor and attachesin_descript to out_descript Programming with Image 9 13 Image 2 1 Library User manual IMAGE structure type descriptor type BINARY_2D Finieerigt hy out_descript the binary image Figure 9 3 IMAGE structure layout after post_op The fixed part of any type descriptor is necessary since it is possible for an IMAGE structure to have type descriptors of different image types simultaneously Because each descriptor contains its type the post_op routine knows which descriptor needs to be destroyed after the operation The fixed data pointer allows the user to use the same macro to find the start of the image data regardless of image type The pre_op function The pre_op function is called with the following arguments int pre_op IMAGE first IMAGE second int mode unsigned long first_spec int second_ident The first two parameters are pointers to the images to be manipulated by pre_op The third parameter tells pre_op what to do The mode be can either COMPARE or ADJUST ADJUST_NIP which are defines from image h The first_spec argument tells pre_op the required image class of the first image More than one class can be specifi
278. ssing the data any temporary changes made by pre_op should be cleaned up and the it must be published that the image data has changed see a interface can display the image This is done by the call to post_op with the image out as the argument This function must be called for any image which has been adjusted using pre_op The value returned by post_op is returned so that the program that called the maximum function can determine whether it completed successfully Error handling and reporting An error can occur in any operation at any time due to several reasons Once an error has occurred the function must try to correct the problem If it is unable to do so it must return gracefully and signal its caller that an error occurred It also must supply information about the nature of the error To determine the location of an error and its nature accurately Image supplies a mechanism to facilitate the error reporting Location of the error To trace the location of an error each function in the Image library participates in a stack mechanism that keeps track of the function that is being executed For this purpose three functions are available void im begin func const char name void im_end_func const char name int im_report_error const char name int val const char info The first statement of a function is a call to im_begin_func with its name as the parameter Just before the function exits in a no erro
279. t be performed in place from an image to that same image they would create re_op and destroy 0ost_op appropriate intermediate images The function post_op also takes care of the automatic display of the result These functions are described in the chapter Programming with Image Now you have made a function that can be put into SCIL_Image but it will only add 1 to images of type g2D which are integer valued If somewhere in other calculations you would have made a real valued image you would not be able to add 1 to it using this function On the other hand you would not want to have an add 1 function for every possible image type you would have to be aware of what image type you would want to process before you could choose the appropriate function SCIL_Image has a way to prevent this you canoverload a command This means that the command itself decides what functions to call dependent on the image types you give as arguments We apply this to extend our example to treat real valued images as well in one overloaded command add_one For that we need to define a function f_add_one which adds 1 to a float image It is very similar to g_add_one see below in the box Getting started 3 31 SCIL_Image 1 4 User Manual We also define one main function add_one which will select either g_add_one or f f_add_one depending on the type of the input image in Its name must correspond to the
280. te standard Windows behavior For instance image processing commands require the name of the image explicitly as an input parameter they do not just work on any image you select with the mouse as you might expect We will point out those non standard properties at the appropriate locations 3 2 Getting started SCIL_Image 1 4 User Manual We will explore these modes of interaction in five sample sessions These sample sessions are not exhaustive but they give you an impression of what can typically be done in each of the interaction modes We end this chapter with an overview of the SCIL_Image commands that can be found in the menus A lot of the factual information of this chapter may be found in the Help pages of SCIL_Image in SCIL_Image press Crtl H go to the Contents menu of the Help facility and press gt gt or select a topic of your choice Session One Viewing Images To start SCIL_Image double click on the SCIL Main Window icon SCIL_Image Figure 3 1 The SCIL_Image program icon If all is well the system displays a menu bar four windows named A B C and D at the top of the screen two text windows named Worksheet and History and a status bar with tool buttons at the bottom If you are using a small screen display some of these windows may not be fully visible Also the image windows will be black at start up in the picture below we have put in an image for added interest
281. ters e Chapter 7 gives a global introduction of the image library e Chapter 8 describes the publish and subscribe message passing mechanism and its use in the image library e Chapter 9 gives an in depth explanation of the image infrastructure and how to program with the image library e Chapter 10 describes the AIO library analysis of images and objects AIO supports facilities for particle measurement for shape and optical density such as in microscopic images of cells e Chapter 11 presents methods for fast morphological binary image processing using bitmapped representations of binary images e Chapter 12 explains how a new image type is created and integrated with the image infrastructure e Chapter 13 demonstrates how to create a stand alone application using the image library by means of a sample application The need for Image2 1 The image library stems from the image processing environment SCIL Image which is a platform independent image processing system SCIL Image includes facilities which lack in the image2 1 library such as the user interface viewing capabilities and facilities for interactive program building For many years it has been a valid approach to build applications within image processing environments such as SCIL Image and run these applications within this system Nowadays we witness the trend in application development shifting from self contained image processing environments towards integ
282. there are some function types that are not part of standard C but particular to SCIL_Image We discuss these now in detail as well as their roles in the CDF Some of the argument types make special use of the min and max fields and others do not use those fields at all Whenever a field is not used a minus sign should be filled in 5 10 Advanced SCIL_Image SCIL_Image 1 4 User Manual text string text differs from the type string A string is a piece of text a text need not be a piece of text it may also contain SCIL variables or values The min and max fields are ignored for this argument type and should contain a minus sign An example scil cdf scilcommand in display_image Command confirm is used for variables that indicate a Yes No choice The min and max fields are ignored An example image cdf UNC set_display_mode ANUAL st_dmode ENU Display ARGS image in A Image choice in NORMAL 0 LIN_STRETCH 4 LOG_STRETCH 8 confirm in Yes Global for 3 D Images switch in Yes Yes No Direct display switch can be used for arguments that can have two values not necessarily Yes or No Internally these values are represented as 1 and 0 but you may give them any name the min field contains text which is used as a symbol for 1 and themax field contains text which is used as a symbol for 0 An example image cdf UNC sobel_diff ANUAL sobe dif ENU Fi
283. this general scheme Acquisition gt Restoration gt Segmentation gt Identification gt Feature extraction gt Selection gt Classification In a number of steps depending on the nature of the problem at hand images are segmented into a binary image with object pixels set to 1 and background pixels set to 0 Component labeling that is assigning an identifying label to a set of connected pixels belonging to the same object results in an identified or labeled image The labeled image is either used to delineate shape features or it is used as a mask while measuring grey value features of objects in the original image Because the same scheme is generally used in many different applications a flexible and generalized framework can improve application development substantially Strikingly enough in many image processing systems all measurements are applied to all the objects in the image and a specification of all requested features needs to be given prior to the analysis The measurements of all objects are then carried out in a single pass through the image and results are printed on the terminal or dumped to a file for further evaluation With this approach it is cumbersome to access and use the measurement results in a program to select or further manipulate objects In addition the single pass approach has a major disadvantage when employing hierarchical classifiers especially in time critical applications When
284. tion header to have the following form type_structure type create char name int type int lenx int leny int lenz The create image function for the byte 2d image type is defined as follows include lt stdlib h gt include lt stdio h gt include image h include im_error h include im_infra h include imtypinf h include roi h include byte_2d h include byte_2dp h BYTE_2D_IMAGE byte_2d_create_image char name int type int lenx int leny int lenz BYTE_2D_IMAGE bp im_begin_func byte_2d_create_image 12 4 New image types Image 2 1 Library User manual if bp New BYTE_2D_IMAGE NULL im_report_error byte_2d_create_image I couldn t allocate BYTE_2D image return NULL if lenz 1 image_output IMO_OUTPUT byte_2d_create_image lenz d overruled set at 1 n lenz lenz 1 Byte2dImageType bp type Byte2dImageData bp BYTE calloc lenx long leny lenz sizeof BYTE if Byte2dImageData bp free bp im_report_error byte_2d_create_image IE_NOMEM byte_2d_create_image unable to allocate image data return NULL Byte2dImageWidth bp lenx Byte2dImageHeight bp leny im_end_func byte_2d_create_image return bp The diagram below shows what the function should create pointer BYTE_2D_I
285. to the on line Help facility You should do this by following the instructions on the Help page User defined Help Pages It requires the editing in Rich Text Format of the file user_def rtf located in the help directory This file contains a sample manual page that you can use to create your own manual pages For this you need an editor that can read this format After you are done composing your Help page according to the instructions your editor window should look similar to this 3 34 Getting started SCIL_Image 1 4 User Manual Microsoft Word user_def rtf Iof x W File Edit View Insert Format Tools Table Window Help 8 x Ole Sialy e ssl lt le ala fox Heading 1 7 t 4 Heading 1 f SCIL NAME NAM EY Normal add_one add 1 to an image K add Normal q SCIL SYNOPSI SYNOPSISY ScilCourier include user def h g ScilCourier include image h T Normal q ScilSyntax int add one IMAGEimage h in IMAGEimage h out g Normal q SCIL DESCRIP DESCRIPTION Normal Add 1 to the image in and store the result int the 1mage out J Normal q SCIL RETURN RETURN VALUESY Normal OK 1 7 ifsuccessfulf Normal NOT_OE 0 ifnot successful Normal q SCIL AUTHOR AUTHOR Normal written by M Y Self 19970204 Normal q SCIL ENDREF far Footnotes Close Footnote Text add_onef F
286. togram PuDNSHeS ieni er a states palin EAE E R Ea 8 8 Error stack publishes osre n e a a E R ai 8 9 Chapter 9 Programming with Image essseossoessoessoesssesssesssosssoossoesssese 9 Intr d u ction to Ma e a A E A E ER S 9 2 Th Image types sst a MR a a NE aN ce E a i E 9 4 Grey val ed IMA pS saae an aeaiee iie ieies 9 4 Binary bitmapped tages nsien hec eres ata a a i aS 9 4 Fl ating point Images cece aia epa a aeiio ee eaatees 9 4 Complex IMA BOS ie ose ie tea aaa a e A EEE EEE N EER EEE ASG 9 5 Labeled Image Suas e N E S A E N TA 9 5 COLOr MARES naeia a EE A E a E a 9 5 The IMAGE SUC DANS iea a a e Vaid a a aao veda ai 9 6 Image FACS atin Onis ise al E EE E ania ihe RENAE a is 9 7 Region of interest data structure rectangular eessseseeeesesreeeesersrrsrresresreseresee 9 8 Region of interest data structure arbitrary shaped sssssesesesssessseseessresses 9 9 Operation COU IST eari ean a es es sane a a E as 9 9 Image DANE yaa ie e a ale ada dae E EA E E Mace dy eee aaa 9 10 Image Color Lo k p Tables nnana aac taga cies S e a ARNT SEn 9 11 Dynamic adjustment Pre_op Post_op e ssssesssesssesesesesesressrseresressrseresrersessresressesee 9 12 The pre op FUNC ON oisin aan Maines eI a a aee e Aa T 9 14 COMPARE O06 set aise sa ni e ii ii a Tete BA eee E NEREA 9 15 ADJUST NIP Moder yogns inonsan aenn ea i a abs 9 15 Output egual to INP t snin i a e aR A EE EE ada ete 9 16 O tp t f specifie ty Pese en anne a a
287. tual conversion of the data are overloaded for each image type In Overruling the default implementation on page 9 31 we explained that overloaded functions can be overruled by the user By overruling the type specific functions for the convert operation the user can implement different conversion methods for certain image types All that needs to be done is to write a function that meets the specifications for the source or destination function described above The new function can overrule the default function for convert by overloading the user s function for the messages conv_to_common for the source function of an image type or conv_from_common for the destination function of an image type In the standard overload files these messages and the default functions can be found just below the obligated service functions Var_objects Var_objects in Image are used to store non image data which is handled by the user They prevent the user from having to manage data arrays and addresses thereof in the interpreter or dialogue boxes To simplify the use of var_objects they are handled like images That is when results are stored in them they can be adjusted dynamically to accommodate the data Because the var_objects can be used to store various types of data a simple mechanism is provided to determine which appear in a dialogue box This enables the application programmer to restrict the visible var_objects to those the user may need
288. tween an image and a display window The dimensions of the image and the type of the data stored in the individual pixels depend on how the image was declared or in some cases the operation of which it is the output The display window is a separate entity acting as a viewport on the data with an independent size Also note that SCIL_Image runs on several operating systems As a consequence the handling of the image displays windows can be vary between the different systems In the text these differences are marked with UNIX Macintosh and MS Windows If you are in doubt about whether a certain operations is valid for the platform you are working on please refer to the on line manual page of that operation When the operation in NOT supported on all systems the PLATFORM section of the manual page will list the systems for which is valid Mouse buttons On most UNIX systems the mouse has 3 buttons which each have a meaning in SCIL_Image On the Macintosh however only one button is available To simulate the MIDDLE en RIGHT mouse button on the Macintosh the mouse button must be used together with the Option key for the MIDDLE mouse button or the Command key for the RIGHT mouse button Although in general MS Windows is used in combination with a two button mouse SCIL_Image for Windows supports all 3 buttons when using a three buttons mouse The title bar of image windows In the title bar of the display window information is displaye
289. type alias MENU menu menu options MANUAL name OPTIONS NOT_COMPILED NO_EXPAND ARGS type lt tab gt type vise default min max prompt FUNC name type alias OBLIGATORY A command entry starts with the key word FUNC followed by the name of the function that defines the command The type of the function may be specified surrounded by double quotes but this is not required Valid types are the standard C types and pointers to them In SCIL_Image it is not possible to refer to defined structures like FILE or IMAGE Ifa function returns a pointer to such a structure char or void should be filled in If no return type is specified the command is assumed to return an int The alias field can be used if the name of the function as it is used in SCIL_Image differs from the actual name of the compiled function This feature should be used rarely if at all since it is a potential source of confusion 5 6 Advanced SCIL_Image SCIL_Image 1 4 User Manual If you wish a menu item containing blanks such as Save As then you should denote this by an here FUNC Save As The actual function name which will be executed should then have an _ at the blank in this example Save_As Of course you could have had the same function called if you had specified the entry FUNC Save_As but then the menu would have shown the underscore which is less elegant
290. type is accomplished by means of function overloading When adding new image types to Image we distinguish two situations 1 A new image type The image type has a data representation different than that of existing image types In this case the steps outlined in Implementing a new image type must be taken 2 A subtype of an existing image type The image type has the same data representation as an already implemented image type but the data is to be interpreted according to the application specific semantics Therefore it will have a specific set of operations associated with it In this case the new image type can make use of functions of the already implemented image types thereby reducing the number of new functions to be written The image types that make use of functions from another image type are referred to as subtypes Implementing a new image type Implementing a new image type in Image means that you must supply a number of required service functions for the new image type namely create_image allocate the data space and fill in the type specific data structures destroy_image free the data space and associated structures copy_part_image copy a rectangular sub image to another image copy_masked_part copy a part masked by a Boolean mask to another image convert_to_common put the image data in a COMMON_LINE convert_from_common read image data from the COMMON_LINE pix_value_str put single pixel data in a string
291. ue patterns to test your algorithms Manipulation contains operations that move pixels around such as reflections and rotations of images Filter contains a standard library of filters both linear for instance the uniform and the laplacian filter and non linear for instance Roberts gradient and the median filter Transform has some transformations Segmentation contains various thresholding algorithms Conversion allows you to convert between different types of images Single_Objects allow you to reduce your processing to single objects in an image for instance to determine their convex hull or to perform specific measurements Statistics has commands that compute some simple image statistics such as mean value minimum and maximum Var_objects contains the manipulation of var_objects which are non image data structures akin to arrays in C Morphology contains many operations from mathematical morphology such as erosion dilation skeleton and hit or miss transforms Some frequently used operations have received special attention in the implementation including the 3x3 erosion and dilation If you are a specialist in mathematical morphology note the options for special points in skeletons and mathematical morphology with arbitrary shaped elements Texture has commands that deal with texture measurements The commands in the remainder of the Image menu can be split into two groups often used
292. ust be in one of the directories specified by the environmental variable SCIL_MAN On the MS Windows platform the manual pages of all functions are integrated in the help file of SCIL_Image the MANUAL keyword therefore is not used on this platform OPTIONS NOT_COMPILED NO_EXPAND OPTIONAL If the OPTIONS field is absent it is assumed that the described function has been compiled If the function has not been compiled then the keyword NOT_COMPILED must be chosen SCIL_Image now knows that it should find the function defining the command in a UFO file name c where name is the name of the function see The Programming Mode Interpreted C functions and UFOs 5 8 Advanced SCIL_Image SCIL_Image 1 4 User Manual Typically SCIL_Image expands any command you may give it which means that it fills in default values of the non specified parameters This may be undesirable for instance if the command can have a variable number of arguments For such commands you should specify the option NO_EXPAND ARGS ctype OBLIGATORY lt tab gt type vis default min max prompt The ARGS field followed by zero or more lines beginning with a lt tab gt character indicates the beginning of the description of the function s arguments This field is obligatory If and only if OPTIONS NO_EXPAND is used see above then you need to specify the arguments by listing them in an ANSI C prototype style e g for the printf
293. ut or output image is actually a ROI The functions provided for these purposes are the pre_op pre operation and post_op post operation functions Each image processing operation created for Image must call pre_op prior to performing any processing on the images and post_op when it is finished As an example consider the threshold function which produces a binary image and assume that the destination image is the input image The pre_op function will handle the creation of an intermediate data structure for output If the image is simply converted to binary then the grey value data will be lost The post__op routine displays the resulting image and cleans up the intermediate data structures Though the conversion steps are admittedly simpler when operations are not performed in place the programmer is freed from being at all concerned with image conversions and image display if the pre_op and post_op utilities are used Further keeping in style with other functions in Image image processing functions should be designed so they can be performed in place as is the case with all functions supported in the library In addition to handling image type and size conversions these utilities also handle regions of interest correctly If a ROI is handed to an image processing function rather than an image the pre_op function will make it possible for the function to treat is as if it were an image The post_op fu
294. utility You have to compile and link it with SCIL_Image Details are given Creating a New Compiled SCIL_Image Version on page 5 16 5 2 Advanced SCIL_Image SCIL_Image 1 4 User Manual e overload c If you want to make a command that accepts arguments of more than one type for instance image processing commands that work both on integer valued and on float valued images you have to make the file overload c which contains this overload information You do this using the program mkoverld started by the make utility You have to compile and link it with SCIL_Image Details are given Creating a New Compiled SCIL_Image Version on page 5 16 Thus if you want to add an interpreted program to SCIL_Image you only need to generate comfile This requires the actions we specified in the sample session Session Four Programming in SCIL_Image on page 3 26 The actions required to add a compiled function to SCIL_Image require the use of a C compiler They are explained in Creating a New Compiled SCIL_Image Version on page 5 16 Making a Command Description File CDF One of the things you must do when adding new commands is create the Command Description File In this section we describe its syntax and how you can express your functional desires in it For easy reference let us repeat the file myfunc cdf from Chapteb UNC myfunc l_yMenu SSCILIMAGE ENU MyMenu Options P
295. value 2D and 3D 16 bit pixel e Binary 2D and 3D in packed format 32 pixels per word e Floating point images single precision 2D and 3D e Complex images 2D and 3D 7 2 Introduction to Image 2 1 Image 2 1 Library User manual e Label images 2D and 3D to facilitate object measurements e Full color images 2D and 3D 24 bits pixel For image input and output the image library supports TIFF JPEG ICS TCL and AIM input only format Advanced and extensive set of image operations The image library supports an extensive set of image operations A large variety of filter and image manipulation routines are available These include arithmetic operations Fourier transforms geometrical routines a versatile image expression evaluator and image measurements Fast implementations Special attention has been given to the implementation of fast algorithms and where possible filters have been decomposed based on linear separable kernels Fast and memory efficient mathematical morphology is available comprising a complete implementation of mathematical morphology using arbitrarily sized and shaped structuring elements Abstract error and I O handling One of the design constraints of the image library was to maintain a strict independence of the user interface and platform This guarantees the possibility to integrate the image processing software with every conceivable user interface and allows to run on a variety of
296. w Levels on page 12 15 that copy the image data to and from an newly allocated piece of memory Although this is not obligatory these lower level routines can be useful for other routines that also might want to copy a part of an image these low levels are listed at the end of this section int byte_2d_copy_part_image IMAGE in IMAGE out int sx int sy int sz 12 6 New image types Image 2 1 Library User manual int width int height int depth int dx int dy int dz BYTE ptr int status im_begin_func byte_2d_copy_part_image if pre_op in in COMPARE BYTE_2D_SPEC ImageTypelIdent in pre_op out out ADJUST ImageTypeSpec out ImageTypelIdent in return im_report_error byte_2d_copy_part_image LE PRE OP p y3 if ptr l_byte_read_part_image ImageInData in SX sy SZ width height 1 ImageWidth in ImageHeight in ImageDepth in return im_report_error byte_2d_copy_part_image IE_NOMEM Unable to allocate part image buffer l1_byte_write_part_image ImageOutData out ptr dx dy dz width height 1 ImageWidth out ImageHeight out ImageDepth out free ptr if status post_op out IE_OK return im_report_error byte_2d_copy_part_image status im_end_func byte_2d_copy_part_image return IE_OK The copy_masked_part is responsible for cop
297. with a chessboard pattern ti_block B 16 255 0 The most powerful tool for image data manipulation is the command eval an expression evaluator that handles both GREY_2D and FLOAT_2D images To fill an image with a grey ramp issue the command eval D xx This is one of the simple things that can be done using eval See Expression evaluation on images eval on page 6 19 for a complete description of eval Region of interest ROI processing To successfully extract the information sought in an image it is often necessary to perform a large number of operations Because the time required to perform an operation depends heavily on the size of the image it is useful to restrict the area operated upon to the region containing data which is of interest If the region of interest ROD is significantly smaller than the complete image the execution time can drop dramatically For example if in an image of 256 256 65536 the region of interest is 64 64 4096 big the number of pixels to be processed is a factor 4 4 16 smaller This means that the time required to perform an operation will be reduced by a factor of approximately sixteen In SCIL_Image a ROI in an image can be defined by specifying the location of the top left corner and the size The ROI then will be treated as if it were a normal image For instance the result of an operation on a ROI will result in an image with the same size as the ROI However the sizes of t
298. write_part_image BYTE ip BYTE data int dx int dy int dz int width int height int depth int dw int dh int dd BYT dp sp int Vi ZF Gl im_begin_func 1l_byte_write_part_image for z 0 z lt depth z for y 0 y lt height ytt dp ip ztdz dw dh dyty dw dx sp data z width height y width memcpy dp sp width sizeof BYTE 12 16 New image types Image 2 1 Library User manual im_end_func l_byte_write_part_image return BYTE l_byte_read_masked_part BYTE ip BOOL_MASK mask int sx int sy int sz int width int height int depth int sw int sh int sd BYTE sp dp ptr register int Vy Ze BYTE mp byte_mask int line_offset im_begin_func 1l_byte_read_masked_part line_offset width 1 8 1 if ptr BYTE calloc size_t sizeof BYTE size_t width height depth im_report_error l_byte_read_masked_part IE_NOMEM couldn t allocate memory for buffer return NULL dp ptr for z 0 z lt depth z for y 0 y lt height y sp ip ztsz sw sh syty sw sx mp BoolMaskData mask z height line_offset y line_offset byte_mask 0x80 for x 0 x lt width x if mp amp byte_mask dpt sptt else dp sptt if byte_mask gt gt 1 mpt
299. ws that are now position on top of each other move the them around so you can compare them To remove these histogram windows UNIX click the middle mouse button in the histogram window Mac Option Click MS Windows click the close box It is also often necessary to reduce the noise in a grey valued image Several types of filters are available for this purpose readfile bnoise A percentile AB 334 uniform A C Binary bitmapped images BINARY_2D amp BINARY_ 3D Data representation of binary images Each pixel in a binary images has value 0 or 1 One bit is therefore sufficient for the storage of each pixel In SCIL_Image we store eight binary pixels in each byte so that a binary image requires only a fraction of the space occupied by a grey valued image A border of long words is reserved around each binary image to enable very fast morphological operations for more precise information see Chapter 11 The binary images in SCIL_Image are displayed in black and red on a color screen and in black and white on a monochrome screen where black is 6 12 The Image 2 1 library in SCIL_Image SCIL_Image 1 4 User Manual used for pixels of value 0 and red white is used for pixels of value 1 To create a binary image in B readfile bnoise A threshold A B 128 In window B a red amp black image has appeared Pointing in it with the left mouse button pressed shows that all pixels in the image have value or 0 The title
300. xecute Shift Enter on the keyboard executes a line or selection Enter on the numeric keypad executes a line or selection Ctrl Enter either of them pops up dialog box for a selection We prefer the use of the Enter on the numeric keypad Now type the command readfile trui A lt hit numeric keypad Enter gt Getting started 3 21 SCIL_Image 1 4 User Manual This instruction is not a legal C statement therefore it can not be executed directly However it is first passed through SCIL_Image scommand expander which translates such typed in instructions into legal C statements if possible The instruction is then executed in the same way the printf statement was As the result of the above command an image is read from disk and displayed in image window A If translation into legal C is not possible the system issues an error message For example try mis spelling the command reafdile trui A lt hit Enter gt The system responds with reafdile t gt variable used but not declared The same command readfile trui A can be issued through the menu system as described in the previous sample session Do so now select Image I_O readfile click OK Note that the command appears in the Worksheet as readfile trui A 300 300 You can use both types of interaction command line typing and menu selection intermixed Of course you may have forgotten what the options are the
301. y the dongle into the printer port the outlet port connecting the computer to the printer Please note that you must have choosen the dongle driver item during the installation 5 On Windows95 some settings were added to your autoexec bat file and on Windows NT some settings were added to the system control panel in the Environment Tab You could now start with chapter 3 Getting Started to get acquainted with the workings of SCIL_Image Before you do glance over the rest of this Chapter It discusses the various folders and files and also tells you what is involved in moving them elsewhere section 2 3 Also if you are not quite happy with the number and size of image windows SCIL_Image displays after starting up you can change them by editing the file scilinit section 2 3 Setting up SCIL_Image 2 3 SCIL_Image 1 4 User Manual The SCIL_Image Folder In this section we describe the contents of the SCIL_Image files and directories The scilimag directory contains a number of sub directories CI scilimage14 Demo E docu Help E Images 3 Prog C comfiles example C include G lib G mylib G overload C resource 3 ste_exmp E bin_2d E float_2d G generic grey_2d Demo brochure pdf 5 license scl docu a comfile P readme doc _j Help ls grab32 dll sciimag exe 2 Images cS grabcol dll a
302. ying arbitrarily shaped Regions Of Interest from and to the parent To accomplish this the function is supplied with two more parameters than the copy_part_image function The second parameter of the function is a Boolean mask The bits which are set in the mask indicate which pixels from the rectangle of the ROI in the parent must be copied to the ROI image An extra flag clear is added to indicate what to do with the pixels in the rectangle of the ROI that are not in the Boolean mask This flag is used by pre_op and post_op when copying the ROI data from the parent to the ROI image and back to the parent pre_op calls the function with clear set unequal to zero to ensure that the pixels in the ROI image that are not covered by the Boolean mask are cleared post_op calls the function with clear equal to zero to ensure that the pixels in the parent that are not covered by the Boolean mask keep their value The parameter list and the calls to pre_op must be int type_copy masked _ part IMAGE in BOOL_MASK mask IMAGE out int sx int sy int sz int width int height int depth int dx int dy int dz int clear if pre_op in in COMPARE type_spec ImageTypelIdent in pre_op out out ADJUST ImageTypeSpec out ImageTypelIdent in return NOT_OK New image types 12 7 Image 2 1 Library User manual the actual copying return post_op out The byte_2d_copy_masked_pa
303. you to use the image as an output image The image cannot be altered of size and or type NOT_IN_DIALOG bit 1 integer value 2 signals to the GUI that the image should not be shown in dialog boxes NO_AUTO_POINT bit 2 integer value 4 signals to the GUI that when pointing in the image viewer with the mouse the pixel information should not be shown Programming with Image 9 7 Image 2 1 Library User manual NO_AUTO_DISPLAY bit 3 integer value 8 signals to the GUI that the image should not automatically be displayed on changes to the image All other bits in the field flags are currently not used but reserved for future use Region of interest data structure rectangular When writing image processing routines in Image the programmer does not have to deal with regions of interest ROI The infrastructure deals with ROIs in such a way that if a ROI is specified to be the input or output for an operation it can be treated as a normal image by the routine A ROI is in fact a small image that exists as a separate image only during operation The infrastructure also keeps track of possible conflict between the ROI and its parent In the event that the parent image changes in size and as a result of that the ROI is no longer located entirely inside the parent image an error is generated at the moment the ROI is used in an operation And when the parent is destroyed the ROI also gets destroyed A ROI in Image is defined by

SCIL_Image User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents