Home

HALCON Version 5.2 Getting Started User's Manual

1. Ball bondings 21 best_match 38 BP 27 Break point 17 27 C 5 38 C 5 38 Export of source code 38 C Interface 38 CAVE 1 Chord encoding 41 closing_circle 41 Color images 39 COM 5 Combo box 17 27 Computer Aided Vision Engineering 1 Contour 42 Control data 40 Cross references 19 Data Control 40 Exchange 39 Floating point numbers 40 Iconic 40 Image 40 Integers 40 Pixel 40 Strings 40 Debugging 5 17 19 40 Dilation 41 Domain 40 41 Environment variables LD_LIBRARY_PATH 14 Environment variables 13 48 ARCHITECTURE 14 DISPLAY 15 HALCONEXTENSIONS 15 HALCONIMAGES 14 HALCONROOT 13 doc html reference hdevelop 14 help 14 images 14 license 14 lut 14 ocr 14 HALCONSPY 14 HOME 15 Erosion 41 Example session 21 Extended line description 42 FA 1 Factory automation 1 fast_match 38 Framegrabber 39 Graphical user interface 15 Graphics window 15 17 24 29 Gray value 29 40 Halcon 5 Installation 6 Library 13 Operator 22 Halcon configuration 13 Unix 13 Windows NT 13 HDevelop 13 38 Program 22 Help 18 HORUS 6 Iconic data 40 Iconic variable 24 Image 40 INDEX Color 39 Matrix 40 Multichannel 39 Object 40 Image file 24 Image file formats 39 Binary 39 BMP 39 Gif 39 PCX 39 PNM 39 Sun Raster 39 Tiff 39 XWD 39 Image object 23 Insertion cursor 17 22 Installation 6 8 Keywords 28 Knowledge base 37
2. 32768 Maximum length of strings 1024 characters Not to be confused with image objects since objects can share the underlying gray values 20 CHAPTER 2 GETTING STARTED Chapter 3 HDevelop Example Session In this chapter you will learn how to solve an image analysis problem from scratch During an example session the most important features of HDevelop are going to be introduced You will become familiar with HDevelop s graphical user interface GUI and learn to use it efficiently The example is chosen from the quality control domain The goal is to detect all ball bondings on a die see Fig 3 1 The HDevelop program corresponding to the example session is included in the HALCON package HALCONROOT examples hdevelop Manuals GettingStarted example dev In particular it will be shown e how to construct a HDevelop program with HALCON operators e how to find suitable operators and parameter values e how to handle iconic and control variables how to handle the graphics window how to generate a region of interest ROD how to solve the example problem using morphology how to solve the example problem by pattern matching and how to use control constructs Read this chapter very carefully and you will understand the basic interactive features of HDe velop This will help you to develop your own sophisticated image analysis programs in the future 21 22 CHAPTER 3 HDEVELOP EXAMPLE SESSION Figure
3. Language interface 38 39 License Development 12 Dongle 12 Floating 7 8 11 Key 12 Nodelocked 7 11 Runtime 7 12 Limitations 19 Maximum image size 19 Maximum length of strings 19 Maximum number of channels per im age 19 Maximum number of contour points 19 Maximum number of image matrices in memory 19 Maximum number of objects per pa rameter 19 Maximum number of polygon control points 19 Maximum value for coordinates 19 Loading an image 23 Main window 15 Menu 16 Edit 16 Execute 16 49 File 16 Help 16 Operators 16 Suggestions 16 Visualization 16 Window 16 Tool bar 16 Activate 16 Copy 16 Cut 16 Deactivate 16 Gray histogram info 16 New 16 Open 16 Paste 16 Pixel info 16 Region info 16 Reset 16 Run 16 Save 16 Set parameters 16 Step 16 Stop 16 Zooming 16 median_image 40 Medical image analysis 1 Memory management 40 Multichannel images 39 Object Image 40 Region 41 Tuple 39 40 XLD 42 Online help 19 27 Operating systems 39 Operator data base 37 Operator knowledge base 6 19 28 Operator library 37 Extension 38 Operator window 15 17 22 23 Text field 22 24 27 Packages 15 Pattern matching 39 PC 22 Performance 28 Platforms 39 50 Linux 39 Windows NT 5 7 12 13 SGI Irix 39 Solaris 39 XLD 42 Tru64 UNIX 39 Windows NT 39 Portable applications 5 Program counter 17 22 Program creation 16
4. 44 CHAPTER 4 PHILOSOPHY OF HALCON Bibliography BB82 D H Ballard and C M Brown Computer Vision Prentice Hall Englewood Cliffs NJ 1982 Jai92 A K Jain Fundamentals of Digital Image Processing Prentice Hall International Inc 1992 Rus92 J C Russ The Image Processing Handbook CRC Press Boca Raton Florida 1992 45 46 BIBLIOGRAPHY List of Figures 2 1 3 1 3 2 3 3 3 4 3 5 3 6 3 7 3 8 3 9 3 10 3 11 4 1 4 2 4 3 HALCON architecture aooaa a eee ee ees Ball ondinss ona di r a ARA Re ne e E The read imape operator u kan ser Barren A rectangular region of interest Are es er ee ars The variable window containing iconic and control variables Selecting an operator by a substring of it s name Detecting bright image parts by thresholding Detecting dark image parts inside the board Removing faulty parts of the raw segmentation with an opening operation The final result of the ball detection based on morphology A template for a ball 2 22 oped a ee A he bee The final result of the ball detection based on pattern matching Structure of HALCON image objects a Different forms and sizes of regions eee Example of region processing y E A AA RR BE 47 Index Aerial image interpretation 1 Architecture 6 Area of definition 39 41 area_center 40
5. Having started HDevelop you will see the following windows on your screen e A main interaction window which includes the HDevelop controls in a menu and a tool bar e aprogram window e an operator window e a variable window and e a graphics window They comprise the HDevelop graphical user interface Thus it is very important for you to get accustomed with these units Reading this manual carefully you will not have any problems using HDevelop to solve your image processing tasks In the following sections the HDevelop windows will be explained briefly For more details see Chapter 3 and the HDevelop User s Manual 16 CHAPTER 2 GETTING STARTED 2 6 1 Main window The main window is the central window and the starting point for your work It comprises the creation of a program operator selection editing etc and the session management loading saving etc This main window is identified by the title HDevelop in the window s title bar It also contains a menu bar and a tool bar e Main window menus In these menus you will find the complete HDevelop functionality which is summarized in the following table For more detailed information see the HDevelop User s Manual Contains Commands for File Loading and saving of HDevelop programs and exit ing HDevelop Edit Editing a HDevelop program Execute Running a HDevelop program Visualization Customizing the graphics windows and handling their output beha
6. Program window 15 17 24 Quality control 1 Raw segmentation 29 Re using code 38 reduce_domain 41 Region 41 Form 42 Object 41 Size 42 Region of interest 25 29 39 41 Remote sensing Result visualization 19 ROI 25 39 40 Runlength encoding 41 Session management 16 Subpixel precision 42 Surveillance tasks 1 System suggestions 19 Alternatives 19 Keywords 19 Predecessors 19 References 19 Successors 19 Troubleshooting 17 Tuple 39 40 Tuple object 39 40 Uninstallation 8 Unix 5 11 13 User extensions 15 Variable window 15 17 24 Visual Basic 5 Visualization Images 24 Regions 26 29 Windows 2000 7 INDEX
7. apply a threshold operator to the input image by selecting the operator threshold from Operators gt Segmentation Specify the input image Die using the combo box If you use the default values for the lower and upper threshold all pixels in the input image with gray values larger than 128 will be included in the resulting output region Therefore you could call the corresponding output variable BrightRegion It might be helpful to set the visualization mode Visualization gt Draw to fill In this case you see all the selected image pixels painted in the current drawing color in the graphics window Double click on Die and BrightRegion again to underlay the visualization of the region with the original image see Fig 3 6 To choose another parameter value just double click on the program line containing the cor responding operator change the parameter and press Apply The operator will be executed immediately and its result will be visualized Repeat this process until you are satisfied with the result then press Ok But remember to clear the window or underlay the original image before each new execution of an operator returning a region Otherwise you only see the accumulated visualization of all regions you generate during the parameter tuning 3 7 Finding the Board as ROI Looking at the input image Die you will notice that the surface of the board is quite bright Thus by applying the threshold operator with the default values 128 a
8. have been extracted from an input image The regions became elements of one region tuple If you want to know the center of all regions you simply have to pass this tuple to the operator area_center HALCON then returns a tuple of integer pairs containing the pixel positions of all centers Now that we know how tuples are processed we can have a closer look at the different object classes of iconic data images regions and XLD objects 4 2 2 Image Objects Image objects contain the pixel data for image processing They may be tuple objects containing more than one image object or a single image object Every single image object consists of one domain describing its area of definition and one or more channels containing the gray values of the pixels cf Fig 4 1 The number of channels is not restricted A monochrome image may only have one channel a color image may contain three channels according to the RGB scheme a multisensor image may have several channels The domain of an image object may be of any size and is represented by a region Thus it can have holes or may consist of several unconnected areas see Fig 4 2 The default domain of an image object is the 4 2 HANDLING ICONIC AND CONTROL DATA 41 smallest rectangle enclosing the image It may be changed to any size e g via the operator reduce_domain so that every image has its individual domain Introducing an area of definition for an image has the following advantage all i
9. rectories e g by executing rm rf HALCONROOT After the installation several environment variables need to be set in order for HALCON to work See section 2 5 2 for details 2 4 HALCON Licensing HALCON checks whether the product is licensed to the user with the help of the FLEXIm license manager package from GLOBEtrotter In this section a brief summary of the default procedure to install a license will be given There are two possible licensing schemes e Nodelocked licenses are only valid on one specific host or for a specific hardware don gle e Floating licenses are valid on arbitrary hosts within a local network A license server checks the number of currently active HALCON applications A new application can be started only if the overall number of licenses is not exceeded Well you should really because the filenames are stored as lowercase on the CD only due to the fact that there is no CD format that works for long file names on both Windows and Unix The installation script fixes the file names to the correct case lOThe full documentation of FLEXIm can be obtained from the FLEXIm end user manual available at http www globetrotter com manual htm 12 CHAPTER 2 GETTING STARTED Note that there are runtime licenses available as well From a user s viewpoint they are treated exactly the same as the ordinary development licenses In order to obtain a license for HALCON from MVTec or your local vendor you
10. In addition under Windows NT HALCON has a COM interface This allows you to use HALCON operators from Visual Basic for example Furthermore it is very easy to integrate your own specific operators into the system The HALCON data structures for iconic data provide a very convenient handling of images re gions and contours Optimized algorithms allow high performance even on standard hardware 6 CHAPTER 2 GETTING STARTED The HALCON system also includes the Computer Aided Vision Engineering CAVE tool HDe velop to speed up the design of image analysis programs A smart online help based on an operator knowledge base provides information about each HALCON operator e g reasonable successors possible alternatives cross references etc See Fig 2 1 for an overview of the HALCON architecture Applications Visual MS Basic Java ein cis Frame grabber Interface Extension Package Iconic Data Management Operator Knowledge Base gt Figure 2 1 The HALCON architecture 2 2 How is HALCON Related to HORUS HALCON is the commercial successor of the image analysis system HORUS which was de veloped at the Technische Universit t M nchen from 1988 to 1996 for Unix systems only Al though MVTec took over the basic concepts of the system all the major components have been revised Furthermore HALCON provides many new features and operators compared to the last HORUS version 4 11 Existing HORUS applicat
11. allocated memory block The HALCON image data base provides several services for object handling to allow a fast and transparent access to image data Images are held in memory To reduce memory costs data is shared by different objects whenever possible The next section gives a more detailed description of the object hierarchy and the handling of image data as well as of control data 4 2 Handling Iconic and Control Data HALCON distinguishes two types of data iconic and control data Iconic data is related to the original image images regions XLD objects see below whereas control data are all kinds of numerical values such as integers floating point numbers or strings Both iconic and control data are processed according to the tuple scheme within HALCON 4 2 1 Tuple Processing Whenever an input or output parameter of a HALCON operator contains data it may be a single object value or a tuple of objects values If a tuple is passed HALCON processes the operator simultaneously on all tuple members For example if you want to process a median filter on six different images you might call median_image six times with varying input images Or you might generate a tuple containing all six images and call median_image once with the tuple as input object HALCON filters all tuple elements simultaneously and returns a tuple containing six filtered images We have seen another example before in chapter 3 where several regions
12. environment variable As a rule it contains several directory names separated by colons on Unix sys tems and semicolons on Windows NT systems e ARCHITECTURE Executable HALCON programs reside in HALCONROOT bin ARCHITECTURE To indicate shared libraries the directory HALCONROOT 1ib ARCHITECTURE is used ARCHITECTURE designates the used platform by an abbreviation e g i586 unknown linux2 2 0 syntax processor hardware_vendor operating system or does not have any content i e it is empty or undefined if the installation for HALCON was made for one architecture only It is useful to include the path HALCONROOT bin ARCHITECTURE or HALCONROOT bin in the path environment variable PATH of a shell script e LD_LIBRARY_PATH The HALCON library path should be included in the system variable LD_LIBRARY_PATH to access shared libraries e HALCONSPY If this environment variable is defined regardless of the value before you start a HAL CON program the HALCON spy tool is activated This corresponds to execute the HAL CON operator set_spy mode on ina HALCON program By this it is possible to monitor an already linked HALCON program during runtime without modifications Additionally you may specify any monitoring mode for the HALCON debugging tool For this you choose a parameter value e g operator input_control output_control etc and activate it This is done by adding these parameters to the envitonment variable H
13. have to perform the following steps e Select your license server node or the computer of your choice for a nodelocked license License server nodes are machines on a network which keep track of how many licenses of HALCON are currently in use They need not be a server in the sense that they have to be a big server machine or have a special kind of operating system e g Windows NT Server installed If you wanttorun HALCON on a stand alone computer the license server must be this computer If you run HALCON on a network any computer on the network can serve as a license server e Get the hostid of the selected host This can be done by running the shipped pro gram lmhostid from the shell Unix or the shipped program 1mutil with parameter Imhostid from the DOS command prompt Windows NT An example output of lmhostid might look like this on a Windows NT system gt lmutil lmhostid lmhostid Copyright C 1989 1997 Globetrotter Software Inc The FLEX1m host ID of this machine is 6909b8b2 You must supply your local vendor with the hostid thus obtained in this example the string 6909b8b2 and the hostname of this machine in your network halconserver for example If the program returns ffffffff this usually indicates that you do not have a network board or on Windows NT systems that you do not have the NetBEUI and TCP IP protocols installed for the network board If this is the case please install these two protoco
14. html within the HALCON package for the latest information 2 5 2 Unix Version The HALCON library and the interactive tool HDevelop use some environment variables which are going to be described in the following It is reasonable to include them in a login script or a shell resource script like cshrc e HALCONROOT This is the most important environment variable It designates the directory where HAL CON is installed Typical paths are for example usr local halcon or usr halcon According to this variable the system switches to subdirectories which are important for running HALCON Some of them are See also page 8 for troubleshooting CHAPTER 2 GETTING STARTED HALCONROOT help The operator data base is situated in this directory It is accessible by all HALCON programs to obtain information about HALCON operators HALCONROOT doc html reference hdevelop HDevelop uses this directory for online help which can be displayed by a suitable HTML browser like Netscape HALCONROOT 1lut User defined look up tables are situated in this directory HALCONROOT ocr This directory includes trained fonts HALCONROOT license This directory contains the license keys necessary for using HALCON HALCONROOT images If the variable HALCONIMAGES see below is not set the system looks for image files in this directory e HALCONIMAGES To search for image files specified by a relative path the system uses this
15. message are No files HALCONROOT help if the package name is system or no help files in one of the user packages Wrong HALCONROOT if the package name is system Check the file protection Probably HDevelop cannot access important files e Help file for package lt package name gt is corrupt Possible reasons for this error message are Inconsistent version of HALCONROOT help if the package name is system or of the help files of one of the user packages Wrong HALCONROOT if the package name is system e Can t open display If you see an error message like this you may have a wrong system variable DISPLAY and or your program is not allowed to open a window by the specified X server 2 8 Help During a HDevelop Session HDevelop offers you a broad range of support while interacting with HDevelop and while pro cessing image analysis tasks 2 9 LIMITATIONS 19 2 9 System suggestions Based on the operator knowledge base the system offers specific advice to choose an appropriate operator Thus there are suggestions for reasonable predecessors and suc cessors If you are searching for different ways of solving an image analysis problem you may have a look at alternatives or references of an operator If you only have some keywords or thematic aspects in mind which an operator has to fulfill but you cannot re member the operator s name you have the possibility to obtain a
16. points in a skeleton junctions_skeleton you have to compute the skeleton itself skeleton e Suggestions gt Successor On the other hand there are typical sequences of operators Hence HDevelop suggests some reasonable operators to be executed following the operator you are currently looking at For example thresholding might be a good idea after applying an edge filter In our example you have just created the input image Naturally this is a very unspecific sit uation with lots of different opportunities to continue However we will accept one of the successor suggestions the threshold operator later on 3 6 Finding the Right Parameter Values Once you have selected an operator from the menu by typing it s name or by double clicking on a program line in the program window it is displayed in the operator window Now you 3 7 FINDING THE BOARD AS ROI 29 have to specify the parameters for the operator Of course you can type in every parameter manually either a control value or a variable name In many cases it is more convenient to use the combo boxes click on the arrow at the right side of the parameter field instead which provide a listing of meaningful values or names of variables you specified within your program Furthermore for control input parameters default values are provided by HDevelop Figure 3 6 The region object BrightRegion containing all pixels with gray values larger or equal 128 For example
17. revision of the man ual has been done e g concerning the installation or the licensing section in chapter Getting Started The example session has not changed much The major difference in handling HDevelop in this basic example is that when changing the visualization param eters the iconic object has to be displayed before the parameters are changed because the last iconic object is automatically re displayed e 2 1 Edition November 1997 The visualization capabilities of HDevelop have been greatly enhanced Zooming into an image can now be handled very conveniently for example CHAPTER 1 ABOUT THIS MANUAL Chapter 2 Getting Started This chapter provides a short overview of the HALCON system It explains the installation and configuration process and it introduces the CAVE tool HDevelop The following aspects are discussed e Whatis HALCON e How is HALCON related to HORUS e HALCON installation e HALCON configuration e The main HDevelop windows and their functionality e Troubleshooting e System help e System limitations 2 1 What is HALCON HALCON is a sophisticated image analysis package designed for product development re search and education It provides a unique library with more than 850 operators covering a wide range of image analysis tasks data visualization and a comfortable debugging mecha nism HALCON operators can be used within C and C programs on Windows NT and many Unix platforms
18. the SERVER line the last entry Finally depending on where you have installed HALCON you may have to customize the path to the MVTec daemon mvtecd Normally this will be HALCONROOT AFLEX1mx ARCHITECTURE Amvtecd However since you cannot use environment variables in the license file you must supply this path manually Paths in cluding blanks are not handled properly under Windows Thus if you have installed HALCON to a directory like C Program Files MVTec Halcon you will have to use the short path C Progra 1 MVTec Halcon instead e Finally for floating licenses make sure that the license manager daemon Imgrd is started at boot time On Unix systems this can be done by starting 1mgrd from the appropriate startup file usually called etc rc boot etc rc local or etc localrc Add the following lines to this file export HALCONROOT usr local halcon etc lmgrd c HALCONROOT license license dat gt etc license log 2 gt amp 1 On Windows NT and Windows 2000 systems the setup program installs the license manager by default as the system service Halcon Licenses which will be started auto matically at boot time 2 5 HALCON Configuration 2 5 1 Windows NT Version All of the work is done by the installation program Setup exe provided The environment variables HALCONROOT HALCONIMAGES and HALCONEXTENSIONS are basically the same as in the Unix Version see section 2 5 2 for details Please look for the file release_notes
19. 3 1 The image to be analyzed All ball bondings on a die have to be detected 3 1 Editing a HDevelop Program After starting the interactive tool HDevelop the GUI as described in chapter 2 6 appears Now HDevelop waits for your input To get your job done you have to create and to process variables either iconic or control by executing suitable HALCON or HDevelop operators The program you develop during a HDevelop session is listed in the program window To add a new line 1 e anew operator you have to perform two steps e Position the insertion cursor on the appropriate line of the program where you would like to insert the new line This cursor is placed by pressing the lt Shift gt button on your keyboard and clicking in the left column beside of the program line with the left mouse button e Select the new operator using the Operator menu or type its name in the operator win dow s text field Now the new operator is displayed in the operator window Provide or adapt the input parame ters and if desired change the names of the output variables Then click Enter to insert the operator as a new program line If you click on Ok instead the operator is not only inserted but also executed immediately Finally if you click on Apply the operator is not inserted into the program but only executed This allows you to test the effect of your parameter settings very quickly and conveniently To execute an arbitrary line of a HALCO
20. ALCONROOT images which is also the default search path for image files as long as you do not provide specific search paths in the HALCONIMAGES variable see section 2 5 3 3 The Graphics Window The size of the initial graphics window is 512 x 512 pixels Thus if your image is bigger or smaller than this size it will be displayed distorted You can interactively change the window size just by picking the lower right corner with the mouse but the easiest way to see the undis torted image is to select Visualization gt Size Window gt Original This enlarges or shrinks the window to the size of the most recently displayed image However in this example we are going to open a new window with the proper size because we will need its window handle later on Again there is a interactive way to do this see menu Visualization but we will use the HDevelop operator dev_open_window instead to include this task in the program itself HDevelop operators are handled in the same way as HALCON operators You can access them via the menu Operators gt HDevelop or directly via the operator window s text field The parameter Width and Height specify the size of the window If you do not know the size of the current image you just specify 1 for both parameters In that case the size of the most recently created image will be used dev_open_window 0 0 1 1 black WindowID Click on Ok and a new graphics window appears It has the same size a
21. ALCONSPY setenv HALCONSPY operator on input_control on 2 6 A FIRST LOOK AT HDEVELOP 15 For further information concerning set_spy please refer to the HALCON reference man ual e HALCONEXTENSIONS This is a list of directories in which user defined extension operators so called packages are kept Each package consists of a number of operators linked into a shared library plus the additional operator documentation in help files and HTML files See the C Interface Programmer s Manual for details on user packages e DISPLAY The system uses this environment variable to open windows It is used in the same way as for other X applications e HOME If you start HDevelop the system is looking for a startup file in the home directory see user manual for HDevelop The corresponding directory is accessed via the system vari able HOME Typically the definition of these environment variables belongs to a start shell script like cshrc or profile Examples for entries in cshrc setenv HALCONROOT usr local halcon setenv ARCHITECTURE i586 unknown 1inux2 2 0 setenv PATH HALCONROOT bin ARCHITECTURE setenv LD_LIBRARY_PATH HALCONROOT 1ib ARCHITECTURE setenv HALCONIMAGES HALCONROOT images 2 6 A First Look at HDevelop HDevelop is the CAVE tool of the HALCON system providing an interactive graphical devel opment environment for image analysis programs It offers a very easy access to the HALCON functionality
22. HALCON Version 5 2 Mr MVTec Software GmbH Getting Started User s Manual December 19 2001 Getting started with HALCON Version 5 2 All rights reserved No part of this publication may be reproduced stored in a retrieval system or transmitted in any form or by any means electronic mechanical photocopying recording or otherwise without prior written permission of the publisher Edition 1 July 1997 Edition 2 November 1997 Edition 3 March 1998 Edition 4 April 1999 Edition 4a December 2001 Mii EC Copyright 1997 2001 MVTec Software GmbH Munich Germany mvtec software GmbH Information concerning HALCON http www mvtec com Contents 1 About this Manual 3 12 gt Reallers sm aa Sue goa rer Sew A Del eek Seale See Ss 1 2 Organization of this Manual 2 Con nn l3 Notation aa a ee ae LE A A ee eS 1 4 Additional Sources of Information o LS Release NOLES tae ar Getting Started Zak What ts HALCON gt ed e el nee o ah 2 2 How is HALCON Related to HORUS o 2 3 HALCON Installation 2 a 08 we O e LW LB da 2 3 1 Windows NT and Windows 2000 o 22s WNIK A Eh 24 HALCON Tigensing A A ede a 2 5 HALCON Configuration cicle a da 2 5 1 Windows NT Version 4 oe 2 ae aca 2 352 UDIR Version oo a A ee ek es oe eR ee et 20 AsBirst LGok at HDeyel p A A E A PE PAE ES 2 6 1 Main window aaa A A a BE ee eS 2 6 2 Program Wind
23. Halcon Note that SystemRoot usually is C Winnt Finally delete the HALCON base directory HALCONROOTY For HALCON 5 0 and 5 1 this DLL was called CSUninst d11 The option c is only included if you have chosen an installation with floating licenses 2 4 HALCON LICENSING 11 2 3 2 Unix To install HALCON on Unix systems mount the CD e g on the directory cdrom On most systems you have to have root rights to do so Note that on some systems most notably Linux the default mount entry usually in etc fstab does not allow user to execute programs on a CD In such cases you have to mount the CD explicitly with a command similar to the following mount dev cdrom cdrom To do the installation you must run the install unix shell script located in the top level directory of the CD This script will ask you in which directory you want to install HALCON and about whether to install various components of HALCON Since all directories and files are read only on the CD you may get some warnings about changed permissions if you run the installation script as an ordinary user as root no warnings of this kind should appear Furthermore on some systems you may get complaints about the fact that the CD contains files with file names longer than 32 characters Prior to installing a new HALCON version you should uninstall any old HALCON ver sion Just remove the contents of the HALCON base directory HALCONROOT and all its subdi
24. N program you have to position the program counter PC at this line just click at the left column beside the line Then press the Step icon in the main window tool bar If you press Run instead all subsequent lines will be executed until a breakpoint or a stop command is encountered A single click on a program line selects this line a double click transfers the line back to the operator window The Copy Cut and Paste commands which are also very convenient to edit a HDevelop program will be explained later during this session 3 2 LOADING AN IMAGE 23 Once you have finished the program you can save it and load it again later or you might export it as C code to generate a standalone application see File gt Save as These basic steps will be used many times during the example session So you will have many opportunities to get used to the different methods of editing program specifying parameter values etc For now just continue with applying your first operator to read an image 3 2 Loading an Image Program _ 5 x read_image Die die3 Operator olx Operator read_image Image Die v image object FileName die3 v filename string Ok Enter Apply Cancel Help Figure 3 2 The parameterization of the read_image operator First of all you have to create aHALCON object that contains your input image You may obtain an image in several ways e You can use a framegrabber to ob
25. Regions gt Transformation is suitable threshold DieROI RawSegmentation 0 50 fill_up_shape RawSegmentation Wires area 1 100 Visualize the variables Wires and RawSegmentation then change the drawing color to see the difference Unfortunately the region found so far contains not only the bonding balls but also some darker areas especially the wires connected to the balls Obviously the bonding balls have a circular shape We use this information to separate the bonding balls from the other dark parts by using the HALCON operator opening_circle in Operators gt Morphology gt Region This morphological operation will suppress all parts of the region where the underlying structural element a circle with a specific radius does not fit within the region see Fig 3 8 opening circle Wires BallRegion 15 5 3 8 FINDING BONDING BALLS USING MORPHOLOGY 31 Figure 3 8 Removing faulty parts of the raw segmentation with the opening_circle operation Visualize the variables Wires and BallRegion then change the drawing color to see the effect of this operation Now there is only one artifact left in the segmentation A large dark rectangular shape in which the circular mask for the opening has fitted as well In order to eliminate this error the intermediate result the region BallRegion is seperated into connected regions via the HALCON operator connection connection BallRegion Balls Note the
26. a framegrabber board e HALCON HDevelop The reference manual for all HALCON operators HDevelop version e HALCON C User s Manual How to use the HALCON library in your C programs e HALCON C The reference manual for all HALCON operators C version 1 5 RELEASE NOTES 3 e HALCON C User s Manual How to use the HALCON library in your C programs e HALCON C The reference manual for all HALCON operators C version e HALCON COM User s Manual How to use the HALCON library in your COM programs e g in Visual Basic e HALCON COM The reference manual for all HALCON operators COM version e HALCON C Interface Programmer s Manual How to extend the HALCON system with your own operators Allthese manuals are available as Acrobat Reader pdf documents The reference manuals are available as HTML documents as well For the latest version of the manuals please check http www mvtec com 1 5 Release Notes Please note the latest updates of this manual e Edition 4a HALCON 5 2 4 December 2001 The manual has been adapted to the currently supported architectures e 4 Edition HALCON 5 2 April 1999 The new Apply button is used in the example session The extended menu and tool bars of HDevelop are described in the introduction The installation process is described in more detail for Unix systems e 3 Edition HALCON 5 1 March 1998 The chapter About This Manual has been introduced A complete
27. ar_threshold dual_threshold dyn_threshold tast_threshold histo_to_thresh hysteresis_threshold threshold threshold_sub_pix image object region object number real number real Figure 3 5 Selecting the threshold operator using the substring thresh Again if you already know at least a substring of the operator name you might also just type this string in the operator window s text field If there is an ambiguity you choose the operator inside the combo box see Fig 3 5 For each HALCON operator the HDevelop online help provides an HTML description Just select Help gt HTML to start your default browser You will see the table of contents of all image analysis modules structured in the same way as the Operators menu Simply select an appropriate chapter and continue browsing until you find your operator There are also many cross references according to the relationships between the operators At the end of the main 28 CHAPTER 3 HDEVELOP EXAMPLE SESSION page there is an index of all operators which helps you to jump directly to a specific operator HTML page Furthermore the menu Suggestions gt Keywords offers access to HALCON operators by keywords By selecting this menu item a window is displayed with the available keywords on the left side and the operators related to the selected keywords on the right side If you select a keyword using your mouse you will see the corresponding o
28. containing marks which control the running of a HDevelop program In the initial HDevelop state you will see the program counter PC represented by a green arrow and the insertion cursor Additionally you may set a break point during program debugging see the user HDevelop User s Manual for a detailed explanation The right area contains the program code It is empty in the initial state 2 6 3 Operator Window This window will show you important information about a chosen operator It indicates all parameter names their types and their values It supports you in specifying a parameter value by displaying a default value and alternative values Additionally you may get online help for each operator Another window element is the text field If you are not quite sure about an operator name you may specify a substring in this text field In this case the system is going to search for all operators in its operator base which include this substring They are displayed in acombo box where you may choose the appropriate operator 2 6 4 Variable Window This is the control unit for iconic and or control variables generated during the execution of the program Double click on an iconic variable to display the contained objects For a closer look on control variables simply double click on them 2 6 5 Graphics Window This window is used to display iconic variables The visualization modes can be adapted to your needs with the functions prov
29. data base consists of two parts a library of image processing operators and a knowledge base The library contains more than 900 operators The operators can be used in any combination All HALCON applications e g HDevelop or HALCON C programs make use of the library to perform image processing tasks The operators exhibit a broad range of functionality from simple tasks e g read image from file to complex processes e g estimate state by Kalman filtering Basically they all have a rather lean complexity and are meant to process single steps of a task instead of the whole task So the typical solution of an image analysis problem consists of acombination of several HALCON operators This concept allows much greater flexibility than others based on fewer but more complex operators that are specialized to perform a certain task Specialized operators are only suitable for special tasks and become worthless when the task changes In contrast to this HALCON operators may be used in any combination If the task changes the choice of the used HALCON operator set may change too It is just like the difference between a predefined automaton and a programmable computer The first one may be easier to use but is restricted to a few tasks The second one is more complex to use but can be adapted to changing problems and may perform nearly any task Among the great number of operators there are some that implement the same task by dif
30. difference BallRegion is one region containing all the pixels of the segmentation so far The different parts of this region are not connected but still they form a single region Balls is a variable containing seven seperated HALCON objects one per connected region Set Visualization gt Colored to 12 Now 12 different colors are used cyclically to paint the region objects It is a very convenient feature of the HALCON system that iconic variables can contain not only single objects but also tuples of objects Most HALCON operators accept such tuples as input and perform their task simultanously on all the contained objects Due to this mechanism you won t have to program loops to process objects one by one in most situations Based on the circular shape of balls the elimination of the rectangular region is now very simple Just apply the select_shape operator in Operators gt Regions gt Features with circularity as the selection criterion select_shape Balls FinalBalls circularity and 0 85 1 0 Double click on the variable Die then on FinalBalls to see the selected regions only Note again that Balls contains seven objects from which you select all circular regions with this single operator call Now the ball segmentation is finished In the next section it will be shown how to perform measurements based on the segmentation result But before continuing recall how simple it was to detect the balls and handle al
31. e TemplateID Note that the check for correct shape and area of each ball is implicitly done by the pattern matching In the previous approach this was performed by the opening_circle operator By the way Program lines that are not needed any more can be permanently removed Just select them left mouse button and choose Edit gt Delete Don t be shy there is an Edit gt Undo as well To temporarily deactivate a line use Edit gt Deactivate and Edit gt Activate to activate the line again 3 11 Using Control Constructs HDevelop offers various control statements to construct more complicated programs For a de tailed specification see the HDevelop Users Manual You already have encountered the assign statement in the previous sections We will now use two other control statements for and if to further improve the example program Up to now all matching positions returned by best_match in section 3 10 were used to display a rectangle This might lead to undesirable results if the matching fails for some of the input objects Just decrease MaxError to 5 for example to see what happens For each input object in BallImage which does not contain the template 0 is returned in Row and Column and the maximum average gray value deviation of 255 is returned in Error The correct way to handle these matching errors is to check for all resulting matching positions whether they are valid or not Simultanously the real number of detec
32. e HDevelop Example Session This chapter contains a complete HDevelop session showing how to solve a specific image analysis task from the scratch Every user interaction is explained in detail to introduce the most important aspects of HDevelop s GUI and its functionality e Philosophy of HALCON Read this chapter to obtain background information about HALCON and its functionality Especially it might be helpful to understand the basic iconic data structures and how to handle these objects 1 3 Notation Except of Unix specific sections in this manual file paths and environment variables are printed in the Windows NT convention e g HALCONROOT examples c_interface halconuser to denote the subdirectory halconuser containing an example package within the HALCON base directory referenced by the environment variable HALCONROOT The same expression in Unix convention would look like HALCONROOT examples c_interface halconuser 1 4 Additional Sources of Information For further information you may consult the following manuals e HDevelop User s Manual An introduction to the graphical development environment of the HALCON system e Framegrabber Integration Programmer s Manual A guide on how to integrate a new framegrabber in the HALCON system Note that in some cases you might define new operators using the C Interface instead of using the standard HALCON framegrabber interface in order to exploit specific hardware features of
33. e problem Now imagine that the maximum region size would be restricted to the image size Thus after processing the first step the dilation the re sulting region has to be clipped to the image size see Fig 4 3 a The following erosion would work on a restricted area and therefore result in a much smaller area than the correct processing see Fig 4 3 b This would make the correct extraction of the fin impossible 4 2 4 XLD Objects XLD objects XLD eXtended Line Description describe areas or any closed or open contour In contrast to regions which represent all areas at pixel precision XLD objects provide subpixel precision There are two basic XLD structures contours and polygons See the HALCON C Interface Programmers Guide for internal details XLD objects can also contain additional information about points and lines such as the gradient strength or the local orientation 4 2 HANDLING ICONIC AND CONTROL DATA 43 region after dilation original image background sen l within background region i j j region after clipping region after erosion within to image size clipped region region after sion without clipping O difference between this region 3 and the background region exctracts the fin resulting region is too small correct extraction of the fin is not possible ee Figure 4 3 Example of region processing
34. egion fill should be displayed with Visualization gt Draw Each time you change the visualization the last iconic object that was drawn into the currently active graphics win dow is redisplayed Fig 3 3 Therefore you need only redisplay an object by hand by double clicking it in the variables window if you want to change the display mode of an older object e Interactive creation of an arbitrary ROI HALCON supports arbitrary regions as ROI see Fig 3 3 Just use draw_region from the menu Operators gt Graphics gt Drawing instead of the sequence draw_rectanglei and gen_rectanglel draw_region ROI WindowID To adapt the program in this way you may select the line draw_rectangle1 just click on that line and deactivate it with Edit gt Deactivate Do the same with the line gen_rectanglel e Computation of a suitable ROI The computation of a suitable ROI based on intermediate results of the image analysis is the most challenging but also the most flexible approach In our example we will detect the board itself and use this as a ROI for the ball detection see section 3 7 Once you have generated a ROI you can reduce the image domain to this region with the oper ator reduce_domain form Operators gt Image gt Domain Specify Die as input image and ROI as input region 3 5 FINDING THE RIGHT OPERATOR 21 reduce_domain Die ROI ImageReduced mean_image ImageReduced ImageMean 11 11 The resulting image I
35. ers For example draw the circles you have computed with the HALCON operator disp_circle in Operators gt Graphics gt Output see Fig 3 9 disp_circle WindowID Row Column Radius 3 10 Finding Bonding Balls Using Pattern Matching The morphologic approach detected the desired bonding balls by suppressing spurious parts of the raw segmentation gained by thresholding In this section you will encounter a completely different approach to this detection problem The pattern matching approach can be divided into two steps 1 Creating a template typically during a training phase 2 Searching for this template in the current image Figure 3 10 A template for a ball In our example we will create a template containing a single ball To do this just copy the two program lines you used to generate a rectangular search space section 3 4 34 CHAPTER 3 HDEVELOP EXAMPLE SESSION draw_rectanglel WindowID Row1 Column1 Row2 Column2 gen_rectanglel ROI Row1 Column1 Row2 Column2 To do so select the two program lines copy them with Edit gt Copy position the insertion cursor at the end of the program and select Edit gt Paste Then position the PC on the draw_rectangle1 command press Run and draw a small rectangle containing nothing but one of the balls see Fig 3 10 Then copy the program line for reducing the image domain to the resulting region ROT reduce_domain Die ROI ImageReduced Now you can transfor
36. execute the supplied program reg_halconx as follows reg_halconx atl dl1 After this the HALCON COM library can be registered by executing reg_halconx without arguments Occasionally the FLEXIm license manager service needed for floating licenses is not installed properly by the installation wizard Sometimes you may decide to use floating license after you have installed HALCON In both cases you can install and uninstall the license manager manually In a DOS command prompt type ZHALCONROOT FLEX1m i486 nt4 installs n Halcon Licenses c HALCONROOT license license dat 1 HALCONROOT license license log e A HALCONROOT FLEX1m i486 nt4 lmgrd exe where HALCONROOT is the HALCON base directory you have chosen during the installation You can check the success of this command using the system control panel Services If a new service called Halcon Licenses appears in the list you re done With the same technique you can remove the license manager again FHALCONROOTY FLEX1m i486 nt4 installs r n Halcon Licenses Troubleshooting for the Uninstallation If you had to register the HALCON COM DLL manually as described above you must unregis ter it manually before you can uninstall HALCON To do so open a command prompt change into the directory misc i486 nt4 on the CD e g type E followed by cd misc i486 nt4 and execute the supplied program reg_halconx as follows reg_halconx u 5Unfortunately a limitation of
37. ferent algorithms This allows to vary how fast and precise a task is performed Consider pattern 37 38 CHAPTER 4 PHILOSOPHY OF HALCON matching for example For a task where only the rough positions of matching points are needed but where the results must be returned as fast as possible the operator fast_match may be used In contrast to this best_match is the better choice when needing the exact positions even wtih subpixel accuracy by using a more time consuming algorithm Apart from the library the operator data base provides a knowledge base that contains detailed information about all operators such as their computational complexity a plain text description of their functionality or typical successor predecessor operators This operator description is used for the automatic generation of different language interfaces like C or C different man uals such as the HTML and PDF versions of the reference manual and the online help system Furthermore HDevelop uses the operator description to support the development of image anal ysis programs e g by suggesting appropriate successors of operators More information on the content of the knowledge base can be found in the HALCON C Interface Programmers Guide in the chapter Operator Description HALCON provides a tool HDevelop for interactive development of image analysis programs In most cases developing image analysis programs is a complex and time consumin
38. g task To help the user find the right operators and parameters for a given task HALCON provides the computer aided vision engineering tool HDevelop This eases developing programs in several ways first HDevelop has a graphical user interface that allows an easy intuitive handling of operators and image data Second operators can be processed and combined interactively All results are immediately visualized so that the user can experiment with different operators and parameter values and directly sees the result Moreover HDevelop supports the user by sug gesting appropriate operators and parameter values An online help explains how the operators work and illustrates their usage by examples If you are satisfied with the result you can save the developed program either in a special HDevelop file format to load it again into HDevelop later on or export the program as C source code By exporting the program can be used by other image analysis applications and can be edited and optimized just like any hand written image processing program O HALCON may easily be extended by new operators Although HALCON already contains more than 900 operators for various tasks you may wish to implement a new one e g in order to access a special hardware or an operator using an alternative algorithm To do so HALCON provides the C Interface that allows the integration of new operators implemented in C It contains several predefined routines and
39. ided in the Visualization menu You may open an arbitrary number of graphics windows Each window can be activated interactively or accessed using its specific window ID 2 7 Troubleshooting This section explains some typical errors while starting HDevelop and their reasons e No license found This error message might have several reasons The file HALCONROOT license license dat is missing and or not readable 18 CHAPTER 2 GETTING STARTED Your license is not valid on this machine There are too many licenses active in case of floating licenses e Lost connection to license server Verify if the license server is running You may also check whether your machine is properly connected with the server For this you may need to contact your system admin istrator e No license for this operator The operator which you try to execute belongs to a HALCON module which is not li censed Obtain a new license including this module e Wrong architecture This error indicates that you are using an architecture which is not compatible with your HALCON version Check the ARCHITECTURE variable or try to install the proper HALCON version e hdevelop Command not found Check your system environment variable PATH e lib can t open file Check whether the system variable LD_LIBRARY_PATH is set correctly see 2 5 2 e No help files for package lt package name gt in directory lt directory gt Possible reasons for this error
40. ific directories within this variable manually This section describes the installation from CD The WWW installation is basically the same However the installed file groups have been adapted to limit the bandwidth needed for the download 3For HALCON versions 5 2 and newer only 4For existing HALCON installations newer than version 5 1 only 8 CHAPTER 2 GETTING STARTED If you uninstall a HALCON version 5 0 or 5 1 with floating licenses the FLEXIm license man ager service occasionally is not removed properly To make sure that the service is removed please start aDOS command prompt prior to the uninstallation and type FHALCONROOTY FLEX1m i486 nt4 installs r n Halcon Licenses Please see also the section troubleshooting for the uninstallation on page 8 Troubleshooting for the Installation On some systems you may get a warning message that the HALCON COM interface library halconx d1l failed to self register If this happens the most likely cause for this is that the Microsoft library at1 d11 is missing on your system It should reside in the directory SystemRoot Asystem32 e g C WINNT system32 In this case you must copy this DLL to the directory SystemRoot system32 manually The DLL is located on the CD in the directory misc i486 nt4 After this you must register the library at1 d11 To do so open a command prompt change into the directory misc i486 nt4 on the CD e g type E fol lowed by cd misc i486 nt4 and
41. ions can be adapted to the HALCON sys tem but since the names and interfaces of many operators have been modified some amount of programming is necessary to do that 2 3 HALCON Installation See the file release_notes html within the HALCON package for the latest information Basically most of the work is done by the installation scripts provided The name was changed mainly due to international trademark considerations 2 3 HALCON INSTALLATION 7 2 3 1 Windows NT and Windows 2000 To install HALCON on Windows NT or Windows 2000 systems simply insert the CD This should automatically start the setup program If the setup program doesn t start automatically execute the Setup exe program located in the directories nt intel Intel of the CD The installation wizard allows to specify whether you would like to install e the full HALCON package Compact Custom Typical e the runtime environment only Runtime or e the demo version hdevelop_demo exe Demo This decision influences the license key you will need to run the installed software For the runtime environment a special runtime license is sufficient The demo version does not need any license key at all The licensing of a full HALCON package and the runtime version is described in section 2 4 Furthermore you can choose between an installation for nodelocked or floating licenses Again please refer to section 2 4 for more details Prior to installing a new HALCON ve
42. irst Other possible reasons might be that the whole HALCON directory was moved to another position on the hard disk or the environment variable 4HALCONROOT was changed manually You have to follow the following steps to recover from the error 1 Check if the environment variable HALCONROOT matches the location of your HAL CON directory You can check this via the system control panel System look for Environment or in a DOS command prompt via echo HALCONROOT If the variable contains nothing or the files are in some other location you have to set HALCONROOT manually via the System control panel 2 Check if these two files are present e In your HALCONROOT directory Uninst isu 10 If the 4 CHAPTER 2 GETTING STARTED e In the directory HALCONROOT FLEX1m i486 nt4 HalconUninst d11 Note that by default the Windows NT Explorer doesn t show DLL files unless you ex plicitly tell it to do so Uninst isu has been created during the installation If this file has been removed an automatic uninstallation is not possible HalconUninst dll isa HALCON specific DLL for the uninstallation process If this file has been removed you can get a copy of the file from the directory FLEX1m i486 nt4 on the HALCON CD Check if the registry entry for the uninstaller is set properly Start regedit exe for example from a DOS command prompt and go to HKEY_LOCAL_MACHINE gt SOFTWARE gt Microsoft gt Windows gt Curre
43. l the intermediate iconic objects generated read_image Die die3 threshold Die BrightRegion 128 255 shape_trans BrightRegion ROI rectangle2 reduce_domain Die ROI DieROI 32 CHAPTER 3 HDEVELOP EXAMPLE SESSION threshold DieROI RawSegmentation 0 50 fill_up_shape RawSegmentation Wires area 1 100 opening_circle Wires BallRegion 15 5 connection BallRegion Balls select_shape Balls FinalBalls circularity and 0 85 1 0 3 9 Working with Control Variables So far we have detected the ball bondings within the image Based on this segmentation we can now perform some simple measurements The most obvious measurement is to count the balls via count_obj in Operators D Object gt Information count_obj FinalBalls NumBalls Similiar to the draw_rectangle1 command in section 3 4 you generated a new control variable called NumBalls containing the desired number of balls There are many different HALCON operators computing specific features of regions see Operators gt Regions gt Features Among them one you might try now is smallest_circle This operator determines the posi tion and radius of the smallest circle containing each input region smallest_circle FinalBalls Row Column Radius Note that FinalBalls contains six objects Thus six circles are calculated one for each of the input regions Therefore the output control variables Row Column and Radius do n
44. le for product development research and education It provides operators covering a wide range of applications Fac tory automation quality control remote sensing and aerial image interpretation medical image analysis and surveillance tasks Rapid prototyping is facilitated by the Computer Aided Vision Engineering CAVE tool HDevelop which offers a highly interactive programming environ ment to design image analysis programs This manual provides all necessary information to install and configure HALCON to under stand it s basic philosophy and to get some insight in HDevelop 1 1 Readers This manual is intended for all new users of HALCON It does not assume that you are an expert in image processing Regardless of your skills it is quite easy to work with HALCON using HDevelop Anybody should be able to understand the basic HALCON principles to solve his image analysis problems quickly Nevertheless it is helpful to have an idea about the functionality of graphical user interfaces GUI and about some basic image processing aspects see e g BB82 Ja192 Rus92 1 2 Organization of this Manual This manual is divided into the following parts e Getting Started This chapter provides a short overview of the HALCON system It explains the installa tion and configuration process and it introduces the CAVE tool HDevelop Consult your platform s documentation for general information 2 CHAPTER 1 ABOUT THIS MANUAL
45. ls and repeat the above steps If you have a Windows NT or Windows 2000 system and no network board installed or if your computer is not physically connected to a network you can also use a hardware dongle for licensing In this case please send the number printed on the dongle e g FLEXID 7 b285cb1b to your local vendor or MVTec e Your local vendor will then supply you with a license file which you will have to install on your system Normally this file must be put into the directory HALCONROOT license and be called license dat see the FLEXIm end user manual on how to config ure this location see the next section for an explanation of the environment variable HALCONROOT A license file typically looks like this SERVER halconserver 6909b8b2 744 DAEMON mvtecd C Progra 1 MVTec Halcon FLEX1m i486 nt4 mvtecd FEATURE MVTec_Halcon mvtecd 5 2 01 jan 0000 2 4C72DB2ECOF315624FAE VENDOR_STRING 16383 ck 97 FEATURE MVTec_HDevelop mvtecd 5 2 01 jan 0000 2 OC52EBFE396EC7015CD5 VENDOR_STRING 16383 ck 43 Do not start the program from the Explorer You must use a DOS command prompt 2 5 HALCON CONFIGURATION 13 The lines SERVER and DAEMON are included for floating licenses only The SERVER line tells FLEXIm where to locate the server You may customize the hostname of the server on the SERVER line the second entry in case you decide to change the hostname of that computer in your network You may also customize the port number on
46. m ImageReduced which is the original image restricted to the rectan gular domain you just have specified to a template This done by the HALCON operator create_template in Operators gt Filter gt Match create_template ImageReduced 5 4 sort original TemplatelD To obtain the center of the template which will be needed for visualization purposes later use the following assign statements TemplRow Rowl Row2 2 TemplCol Coll Col2 2 After the creation of the template you have the choice between different matching algorithms We will use a two step strategy First a fast matching operator is applied to find all image points where the template might match In the second step within each cluster of such points the exact position of the template will be determined The HALCON operator fast_match again in Operator gt Filter gt Match delivers the region Matches containing all possible matching positions fast_match Die Matches TemplateID 20 Double click on the original image Die and the result region Match to see these positions The input parameter MaxError is set to 20 in this example There is a trade off between performance and fault tolerance of the matching Decrease the value and watch the execution time when applying the operator For very small values of the tolerated error some of the balls might be missed On the other hand increasing the value decreases the performance significantly Be
47. macros for easy handling of image data and memory objects in C Once a new operator has been successfully integrated it can be used like any other HALCON operator The C Interface Programmers Guide contains detailed information about extending the operator library O HALCON allows continued application of already implemented image processing software Software engineering is a time consuming process and in most cases expensive as well So everyone is interested in using already implemented software as long as possible and no one is bent on re implementing their software with every installation of new software components Therefore HALCON supports the continued application of already implemented software Convert your programs into new HALCON operators to use them within HALCON which can be done via the C Interface as described above Moreover HALCON hardly restricts the hard ware configuration A wide range of machines and systems can be used Various framegrabbers 4 1 FACTS ABOUT HALCON 39 are supported New ones may be easily integrated see below Thus you need not change your hardware configuration and therefore also need not to change your software O HALCON supports a wide range of platforms HALCON is largely architecture independent and therefore allows you to use the system of your choice It supports different platforms such as Solaris Tru64 UNIX SGI Irix Linux or Windows NT O New framegrabbers may easily be i
48. mage operators work only within this image domain This allows to focus the processing on a region of interest The amount of data to work on becomes smaller so that the processing is speeded up An example that explicitly makes use of this can be found as HDevelop program named autobahn dev under the path HALCONROOT examples hdevelop Applications Sequences tuple image object image object 1 image object 2 domain channel 1 channel 2 ee ee l N ha L Y l l J l N IA FE l FE FE runlength gray value gray value encoding matrix matrix Figure 4 1 Structure of HALCON image objects 4 2 3 Region Objects Regions define areas in images e g the domain of an image or the light areas of an image Region objects consist of one or more regions They are handled as separate objects indepen dent from any image object although they may represent an area within a specific image This allows to use flexible shapes and sizes for regions and to store them efficiently Only the region area must be represented in memory independent from the gray values of the pixels More over this makes possible to define overlapping regions for the same image without the implicit merging that would happen when storing regions as matrices Region data is represented by a special variant of the runlength encoding by a chord e
49. mageReduced is only defined within the ROI Clear the window and double click on the variable ImageReduced to see this effect You might apply a couple of operators to the original image and the reduced image to see the increase in performance For example try mean_image from Operators gt Filter gt Smoothing with Die as input image Then clear the window and double click on the mean_image program line to change the input image to ImageReduced To repeat the above sequence with different ROIs just place the PC on draw_rectangle or draw_region depending on which method you have activated and press Run In order to stop the execution of the program before a specific operator is applied place a break point BP in the program line containing this operator Control key and mouse click in the left column 3 5 Finding the Right Operator You have created your first HALCON object the original image But how to continue There are more than 800 operators to choose from Naturally the selection of a suitable operator sequence depends on the image analysis task you have to solve In general it is up to your experience and image analysis knowledge to make the right selection However HDevelop supports you in many ways during this program development The HALCON operators are hierarchically structured Just browse the Operators menu to get a first impression Operator OF x Operator thresh v auto_threshold bin_threshold ch
50. n operator by specifying one or more keywords Online help on operators HALCON offers a detailed online help for each HDevelop operator These manual pages are highly structured thus allowing you to get an understanding of an operator easily To get a survey of the operator s relations cross references to other operators are shown To use this help you only need an appropriate HTML browser Debugging help You may choose between different modes of running a program You have the possibilities to perform single steps of a program to set breakpoints or to run a program continuously To get meaningful debugging support you may also specify the environment variable HALCONSPY see chapter 2 5 2 Result visualization HDevelop offers the immediate inspection of an operator s results This visualization mechanism helps you to spot problems quickly Additionally it is very easy for you to modify graphics output parameters to optimize your graphics results Limitations There are some HALCONIimitations which you have to keep in mind although you won t come even near these limits in most applications Maximum image size 32768 x 32768 Maximum number of image matrices in memory 200 Maximum number of objects per parameter 100000 Maximum number of channels per image 500 Maximum number of values in a tuplet 1000000 Maximum number of contour points 30000 Maximum number of polygon control points 10000 Maximum value for coordinates
51. ncoding for every line chord of a region its row index y coordinate of chord line number and the x coordinates of its start and end point is stored Both the start point and the end point belong to the region This allows a very compact representation of regions in memory and an efficient region processing especially with morpholgic operations Regions may be of any size and therefore even may exceed the image size This is an ad ditional advantage since it prevents artifacts For example imagine you want to close a region with a circular structuring element to smooth the region s boundaries and to close holes within the region e g via the operator closing circle You use this to extract a little fin see Fig 4 3 or cf the HDevelop program example fin2 dev under the path HALCONROOT examples hdevelop Applications FA The closing is done by a dilation followed by an erosion both with the same circular structuring element in our case a circle 42 CHAPTER 4 PHILOSOPHY OF HALCON mi region 1 mi region 3 regions may contain holes regions may be compact Ea region 2 Ea region 4 regions may consist of regions may exceed several unconnected areas the image size Figure 4 2 Different forms and sizes of regions of radius 250 you may think that 250 is rather large but with the advanced performance of HALCON this causes no performanc
52. nd 255 a reasonable raw segmentation of the board can be achieved We will now transform this raw segmenta tion into a region of interest for the ball detection Select the HALCON operator shape_tans with BrightRegion the result of the thresholding as input region and rectangle2 as value for the parameter type to obtain the smallest rectangle of arbitrary orientation containing BrightRegion The resulting region called ROI is a reasonable approximation of the board 30 CHAPTER 3 HDEVELOP EXAMPLE SESSION in the image As in section 3 4 reduce the domain of the original image to this ROI with the reduce_domain operator threshold Die BrightRegion 128 255 shape_trans BrightRegion ROI rectangle2 reduce_domain Die ROI DieROI The following computations will be performed only on the resulting image DieROI within the board 3 8 Finding Bonding Balls Using Morphology in N Figure 3 7 The region object RawSegmentation containing all dark pixels within the board If you look closer at your image you will notice that the bonding balls to detect are darker than the die itself Therefore another thresholding might be suitable as the next operation see Fig 3 7 Due to noise the resulting region all the dark parts within the board might have some small holes which can obstruct further processing For a controlled filling up of these arti facts the HALCON operator fill_up_shape in Operators gt
53. ntVersion gt Uninstall gt Halcon lt Version Number gt where lt Version Number gt is the version number of the HALCON version you want to uninstall or is not present There you should find a key named UninstallString The value of this string should be something like this C WINNT IsUninst exe f C Program Files MVTec Halcon Uninst isu c C Program Files MVTec Halcon FLEX1m i486 nt4 HalconUninst di1 Make sure that the path C Program Files MVTec Halcon or equivalent in the above example points to the HALCONROOT directory Note that filenames that contain blanks have to be quoted with path This is true in the above example since Program Files does contain a blank Older versions of the HALCON setup program failed to do so If you encounter unquoted path names containing blanks please insert the quotation marks yourself Close the registry editor and try to run the uninstaller again automatical uninstallation fails for whatever reason proceed as follows For floating licenses only Uninstall the license manager service by executing FHALCONROOTY FLEX1m i486 nt4 installs r n Halcon Licenses Start regedit and delete the keys HKEY_LOCAL_MACHINE gt SOFTWARE gt Microsoft gt Windows gt CurrentVersion gt Uninstall gt Halcon and HKEY_LOCAL_MACHINE gt SOFTWARE gt MVTec Using e g the NT Explorer delete the directory SystemRoot AProfileslA11 Users Start Menu Programs
54. ntegrated Although HALCON already supports several framegrabbers e g from Imagenation Bitflow or Integral Technologies you may wish to install and use a different one To do so HAL CON provides a transparent framegrabber interface with sample source code for integrating new framegrabbers The only thing you have to do is to adapt this code to a specific framegrab ber and to link it to the HALCON code The Framegrabber Integration Programmers Manual describes in detail how to do this O HALCON supports data exchange with other applications For easy exchange of image data HALCON supports several file formats such as Tiff Sun Raster Gif PNM PCX XWD BMP binary Other data e g integer result values for image coordinates may be exchanged via the HALCON language interface O HALCON processes color and multichannel images HALCON works on all kinds of image data binary monochrome color or multichannel images You may use all these kinds of images in the same manner Multichannel images are useful to work on images that were acquired by a multisensor system O HALCON allows to focus image processing on a region of interest Each image object has a domain its area of definition that can be changed by the user When performing an image operator it is processed only within the domain of the input image This concept allows to focus image processing and therefore to speed it up O HALCON provides an ea
55. ore while setting an image part to zoom in After you finished the drawing the four control output variables Row1 Columni Row2 and Column2 will appear in the variable window see Fig 3 4 To gen erate the corresponding ROI as an HALCON object use the operator gen_rectanglei from the menu Operator gt Regions gt Creation with the variables Row1 Column2 as input parameters You might use ROI as the name for the resulting output object 26 CHAPTER 3 HDEVELOP EXAMPLE SESSION draw_rectanglei WindowID Row1 Columni Row2 Column2 gen_rectanglel ROI Row1 Columni Row2 Column2 By executing this operator you create another type of HALCON object a region see section 4 2 3 Variable Watch Bigg Iconic Variables Control Variables WindowHandle 3604 Row 97 5 Column 295 5 Row2 385 5 Column2 665 5 Figure 3 4 The variable window containing iconic and control variables The default visualization for a region is the shape of the region here the rectangle drawn with the current color into the graphics window You may manipulate this graphics out put by choosing items of the menu Visualization or executing the corresponding HDe velop operator dev_ For example set the drawing color to green with Visualization gt Color increase the line with for drawing to 3 pixels with Visualization gt Line Width and specify that only the region boundary margin instead of the whole r
56. ot contain single values but tuples of values Again no loop construct is necessary to process all objects Tuples of ncontrol values are shown in brackets within the variable window If a tuple variable contains too many elements to be shown at once the value list is truncated and three dots are added For inspecting such a tuple just double click on the variable you want to see with the left mouse button A window will be opened containing all the elements of the tuple Figure 3 9 The final result of the ball detection based on morphology The further processing of control data is facilitated by a built in formula interpreter See the HDevelop User s Manual for details For example you could compute the average value of the tuple Radius with the following HDevelop assign statement see Operator gt Control 3 10 FINDING BONDING BALLS USING PATTERN MATCHING 33 assign meanRadius sum Radius Radius which is displayed in the program window as meanRadius sum Radius Radius A more reasonable measurement concerning the quality control task might be the computation of the minimum area of the balls This can be achieved by applying the HALCON operator area_center in Operators gt Regions gt Features which calculates the area of each ball followed by another assign statement area_center FinalBalls Area RowCenter ColCenter mimArea min Area Of course tuples of control values can also be used as input paramet
57. ouse to the lower right corner of the desired rectangle Release the left button and click the right button once Now you have specified the image part you are interested in All visualizations are now zoomed to this image part This is the interactive way to set an image part Another possibility is the direct input of the coordinates of the image part in Visualization gt Set parameters gt zoom or to use the HDevelop operator dev_set_part To visualize the whole image again click Reset in Visualization gt Set parameters gt zoom 3 4 Generating a Region of Interest ROD Very often it is useful to specify a ROI which is used in the following processing steps By re stricting the image domain the performance can be increased significantly because all following image processing tasks will be performed only within this ROI and not on the whole image Figure 3 3 A rectangular and an arbitrary region of interest ROD To generate a ROI you have the following possibilities e Interactive creation of a rectangular ROI Select the operator draw_rectangle1 from the menu Operators gt Graphics gt Drawing to draw a rectangle in the graphics window Just execute the operator with the default names for the output variables containing the pixel coordinates of the upper left and lower right corner of the rectangle This creates a new line in the program window and HDevelop waits for you to draw the rectangle like you did bef
58. ow 2 5 eas aon a a er Bes de Sod 2 6 3 Operator Window 4 6 64644 844644 644 844844 S445 44 20 4 Variable Window su o e aoe a ee ee ee a 2 6 5 Graphics Window meses 28 200 e a Ree Bs 221 Troubleshooting 4 224 a2 e nl ee Gud des dal de oc de 2 8 Help During a HDevelop Session 2 2 2m nn nennen 2 97 1 Eimitations ee a ee ee ee ee te ee ee HDevelop Example Session 3 1 Editing a HDevelop Program 306 4 a 22 N a ae 3 2 Boadins an Imagen do Brahe Gow ie er ele Seeley ale Y 3 3 The Graphics Window ss A ddl ei hd Book SS AE ERS SOS 3 4 Generating a Region of Interest ROD o o 3 5 Finding the Right Operator e e 3 6 Finding the Right Parameter Values o e 3 7 Finding the BoardasROl o o e 3 8 Finding Bonding Balls Using Morphology 3 9 Working with Control Variables o o e ii 3 10 Finding Bonding Balls Using Pattern Matching 3 11 Using Control Constructs 4 Philosophy of HALCON 4 1 Figures Index Facts about HALCON 4 2 Handling Iconic and Control Data 4 2 1 Tuple Processing 4 2 2 Image Objects 4 2 3 Region Objects 4 2 4 XLD Objects are irre CONTENTS Chapter 1 About this Manual This manual is a quick guide to HALCON the software solution for machine vision appli cations HALCON is a sophisticated image analysis package suitab
59. perators in the right area Each mouse click in the left area marks the keyword and adds some operators in the right area To finally select an operator and display it in the operator window just click on the operator name using your left mouse button Once you have selected one of the operators like read_image in our example you can enquire information related to this operator from the operator knowledge base e The HTML online help Just click on the Help button to launch your default web browser and get the complete online documentation of the selected operator e Suggestions gt Alternatives If an intermediate result does not fulfill your requirements you might want to choose an alternative operator which might be more suitable for your task Note that in many situ ations there is a trade off between performance and accuracy of the results For example you might use the high performance smoothing filter mean_image in one situation and the higher quality smoothing filters gauss_image or smooth_image in another e Suggestions gt See also The online help of HDevelop also offers references to other operators which are not exactly alternatives but related to the current operator in some way For example if you currently look at read_image you might be interested in the write_image operator as well e Suggestions gt Predecessor Many operators require a specific input provided by other operators For example before computing junction
60. rsion you should uninstall the old version This is especially of importance to clean up the registry Please use either e the option that the HALCON setup program will show you if you have old versions installed this is the preferred way since it avoids the problem with the path variable described below e the system control panel Add Remove Programs or e Start gt Programs gt Halcon gt Uninstall Halcon to remove the HALCON package Typically this is all you have to do However please refer to page 8 for trouble shooting if you encounter any problems Note that an uninstallation will also remove the example images and the documentation You might want to copy these directories to another directory in order to prevent this However we do not recommend this strategy In most cases there will be both additional images and a revised documentation included in a new HALCON release Thus these old directories will be out of date anyway Please note that the uninstallation of HALCON versions 5 0 and 5 1 will delete the envi ronment variable PATH This is due to an error in the third party installation software that is used If you have installed other software packages that have modified this variable please pre serve the old settings prior to the uninstallation Just use the system control panel System entry Environment to rename the variable e g to PATH_ After the uninstallation rename it again back to PATH and remove the HALCON spec
61. s your image But the new graphics display remains dark It is now the active graphics window and waits for your input It is up to you what to display in this window Double click on the iconic variable Die in the variable window by using the left mouse button in order to display this image Now you see the input image without distortion see Fig 3 1 The default visualization for an image is to display the image matrix i e the gray or color value of each pixel You can easily switch to other visualization modes via Visualization gt Set parameters gt paint Choose for example 3D plot and select Update to see a 3D plot of the image The various visualization modes are desrcipted in detail in the HDevelop Users Manual Do not forget to change the mode back to default before continuing with the example session All the visualization modes can be modified within your program as well For example the display of the histogram is activated by dev_set_paint 3D plot See also section 3 4 to learn how to manipulate the visualization of regions 3 4 GENERATING A REGION OF INTEREST ROI 25 Furthermore it is possible to zoom into the image Just select the menu item Visualization gt Set parameters gt zoom Click on the upper Interactive button Now you can specify a rectangle within your graphics window To do this you press the left mouse button to indicate the upper left corner Keep the button pressed and drag the m
62. sides fast_match there is also a HALCON operator fast_match_mg working on different resolution levels which in general increases the performance Just try fast_match_mg Die Matches TemplateID 20 3 to perform a search with three resolution levels and compare the execution time with the one resolution method underlying fast_match After we have derived all posible matching positions we decompose the corresponding region into connected components connection Matches BallROI If you specified a reasonable template and a suitable MaxError this should result in six regions corresponding to the six fully visible balls in the image 3 10 FINDING BONDING BALLS USING PATTERN MATCHING 35 count_obj BallROI NumBalls Within each of the regions in BallROI we would like to determine the exact position of the template This can be done by another HALCON operator called best_match BallROI contains only regions Thus we have to add an image that is gray values to the objects prior to the final matching This is done with the operator add_channels in Operators D Image gt Channel add_channels BallROI Die BallImage best_match BallImage TemplateID 20 true Row Column Error Note that the best_match operator is applied to all objects contained in BallImage Thus the control output variables Row and Column specifying the exact matching position and Error containing the matching error should be tuples of values To
63. sy handling of tuple objects and processes tuple elements simulta neously Tuples are very helpful in many areas of application since they enable you to work with sets of images regions or control data With HALCON tuples of objects or control values can be handled just as one object Operators can either work on a single or on a tuple object You need not bother about the amount of elements within a tuple You simply pass one variable containing the tuple to an operator HALCON decomposes it and works simultaneously on the tuple elements O HALCON implements efficient fast region processing This speeds up the processing and eases the handling of regions Regions may overlap and can be of any size They are stored by an optimized encoding to reduce memory costs O HALCON provides very fast pattern matching operators Pattern matching is very helpful in numerous areas of application but often not used due to its costs of computation time HALCON allows to use pattern matching where you want to as it uses a very efficient implementation of various pattern matching algorithms 40 CHAPTER 4 PHILOSOPHY OF HALCON O HALCON is based on an efficient management of image and data objects The HALCON memory manager uses fast efficient algorithms to speed up image processing It provides mechanisms for controlled memory accesses and services for debugging such as checking with every deallocation whether any write access exceeded the
64. tain an online image see menu item Operators gt Image gt Framegrabber e You can read an image from a file see menu item Operators gt File gt Images and menu item File gt Read image e You can create a new image see menu item Operators gt Image gt Creation In the example we will use the read_image operator Just select read_image in the submenu Operators gt File gt Images or type read_image in the operator window s text field and press Enter Now the operator should be displayed in the operator window see Fig 3 2 The first parameter is the image object you are about to create Change the default name of the 24 CHAPTER 3 HDEVELOP EXAMPLE SESSION corresponding variable to Die You get a text field focus by clicking the left mouse button inside the desired text field The second parameter is the name of the image file Change it to die3 read_image Die die3 Having specified all parameters you should press Ok to transfer the operator to the program win dow and to execute it Now you have created the first line of your program in the program win dow The variable Die containing the image object you just created is displayed in the variable window Simultaneously the result of the operator in this case the original image is displayed in the graphics window If you encounter an error trying to load the image check whether your environment is set up correctly The image should be located in H
65. ted balls can be computed Initialize a counter with the assign statement RealNumBalls 0 Then program a loop with the counter i and the end value NumBalls control statement for Inside the loop check the error value of each match and skip faulty matches for i 1 to NumBalls by 1 if Error i 1 lt 255 disp_rectangle1 Row1 Row i 1 TemplRow Columni Column i 1 TemplCol Row2 Row i 1 TemplRow Column2 Column i 1 Temp1Col RealNumBalls RealNumBalls 1 endif endfor Display the original image Die again and execute the above loop to see the result 3A template position which is actually outside of the current image because Row and Column specify the position of the center of the template Chapter 4 Philosophy of HALCON HALCON is more than just a library of image processing operators or another graphical user interface for image processing It is a comprehensive system for machine vision applications suitable for product development research and education It also includes a powerful computer aided vision engineering tool called HDevelop It might be helpful to have a closer look at the philosophy of HALCON to find out more about the different tools and features 4 1 Facts about HALCON Let s start with some facts describing the main characteristics of HALCON O A comprehensive operator data base supports efficient flexible image processing and eases the development of image analysis programs The operator
66. the third party installation software prevents us from installing the library from the HALCON setup program The following is one long command line 2 3 HALCON INSTALLATION 9 As mentioned above if you uninstall a HALCON 5 0 or 5 1 version with floating licenses the FLEXIm license manager service occasionally is not removed properly To make sure that the service is removed please start a DOS command prompt prior to the uninstallation and type HALCONROOT AFLEX1m11i486 nt4installs r n Halcon Licenses Note that the DOS command prompt cannot handle paths including blanks properly Thus if you have installed HALCON to a directory like C Program Files MVTec Halcon you will have to type either C Progra 1 MVTec Halcon FLEX1m i486 nt4 installs r n Halcon Licenses or quote the command as suggested here C Program Files MVTec Halcon FLEX1m i486 nt4 installs r n Halcon Licenses Alternatively you can also change the directory via cd HALCONROOT FLEX1m i486 nt4 which is possible even without quotes and type installs r n Halcon Licenses afterwards Occasionally a message like Internal Error unable to load or call external DLL Please contact your ven dor for more information is printed when the uninstaller is invoked and the uninstall process is terminated This prob lem is most likely to appear when anew HALCON version was installed over an existing one without completely removing the old files f
67. viour Operators Submenus containing programming constructs HDe velop operators and all HALCON operators Suggestions Support choosing an operator Window Window management Help Opening online documentation and guides e Main window tool bar This tool bar offers shortcuts for frequently used commands Furthermore it provides the means to control the execution of an HDevelop program Performs this Operation Deletes the current program in the program window Opens a new HDevelop program Saves a HDevelop program Deletes program lines that have been selected in the program window Copies program lines that have been selected in the program window Inserts copied lines in the program window Executes the HDevelop program Executes the next command of the HDevelop program Stops the running HDevelop program Activate Activates the lines selected in the program i e makes them executable Deactivate Deactivates the lines selected in the program Reset Resets the program into its initial state and clears all variables Pixel info Opens the pixel info window Zooming Opens the online zooming window Gray histogram info Opens the interactive gray histogram inspection window Region info Opens the interactive region feature inspection window 2 7 TROUBLESHOOTING 17 2 6 2 Program Window This window is used to display a HDevelop program This might be an entire program which has been loaded or single operators only It has two parts a left area
68. visualize the results we display rectangles like the one used for the template generation shifted by the offset beween the matching positions and the center of the template see Fig 3 11 disp_rectangle1 WindowID Row1 Row TemplRow Columni Column TemplCol Row2 Row TemplRow Column2 Column TemplCol You should make sure that the region visualization is set to margin using Visualization gt draw before executing this command Graphics Window Figure 3 11 The final result of the ball detection based on pattern matching If you do not need the template any longer it should be cleared with the HALCON operator clear_template Summarizing the ball detection using pattern matching was done by the following sequence of HALCON operators read_image Die die3 draw_rectangle1 WindowID Row1 Columni Row2 Column2 gen_rectanglel ROI Row1 Columni Row2 Column2 reduce_domain Die ROI ImageReduced create_template ImageReduced 5 4 sort original TemplateID Again there is a multi grid version of this matching operator but since we already restricted the search to a very small part of the image it is not too interesting to use best_match_mg in this case 36 CHAPTER 3 HDEVELOP EXAMPLE SESSION fast_match_mg Die Matches TemplateID 20 3 connection Matches BallROI add_channels BallROI Die BallImage best_match BallImage TemplateID 20 true Row Column Error clear_templat

HALCON Version 5.2 Getting Started User's Manual

Contents

Download Pdf Manuals

Related Search

Related Contents