medeaLAB Count & Classify Image Analysis System
Contents
1. Changes of the parameters above must be confirmed with the button Modify Color Display color of ROT Change Position Move ROI using the arrow buttons You may also use the arrow keays of the keyboard If Select Active is checked only the active highlighted ROI will be moved If Select All is checked all ROIs will be moved simultaneously Add off on Using the Add mode you may create new elliptic ROIs of the specified size in the image by simple clicking the left mouse button This is useful e g to mark multiple positions for inspection tasks Click the Add on off Button to toggle the Add mode the button displays the current state 91 IMAGING SYSTEM 6 Methods for preprocessing image data 6 1 Digital filters Apply a filter directly on the active image window by using the menu Image Filters Note that filter operations are executed only in the active ROI You may choose from various filter methods for image enhancement Soften Sharpen Sobel Median Dilate Erode Inverse Brighter Darker Remove Marked Pixels Deinterlace Mirror Flip Binarize Edge Lines Squares Color Channel Settings Selection Run selected The menu command Settings opens the Filter Settings dialog which allows you to determine e g the kernel width of the filter operation and the number of iterations Kernel idth Iterations Canny Threshold 50 Canny Threshold 200 The Canny Thr2. Counting coherent objects using blob analysis Blob analysis will mark all object pixels and is suitable for measuring e g the total object area Start the measurement using the menu command Image Search Blobs The results will be displayed in a table E 823 JPG W Area Measurement Sel index osx Posy arealam R Green ue 1 62 1688 00 Results of the Search Blobs function 7 3 Measuring area fractions within ROIs This method counts all object pixels within thresholds in relation to ROI area This is useful for e g quantitative fluorescence evaluations First define one or more ROIs chapter 5 then start measurement using the menu command Image Integration ROIs The results for all ROIs will be displayed in a table 97 2 Bad Fe tt hoe BS OE Ee ee id ee een LH ne T ehh hhh ERTL TUE SAN eS he TR ta a gt K P PA f Pa CPP P PSL RA EEE EEE LEE TEED SOOO EREET Fprrrrtttttttt a an 2 fig wer u a gt hal vA E AO Measurement ROI Area um Red Green Blue Homogenty Object 1 949683 00 59 184 60 1 00 2 990996 00 47 114 28 1 00 18 0 gt 49923 00 32 90 25 1 00 10 0 ROI Measurement three freehand ROls to determine the area rate of bright fiber material Object in different samples 7 4 Length measurement For length measurements of filamentous objects e g roots fibers two different methods are available in medeaLAB Count amp Classify It
3. File Camera WDM HIL Digilam Prosilica GigE Currently the following video sources are supported by medeaLAB besides image files and TWAIN drivers Video file File in AVI format Camera WDM Digital video camera with Microsoft Windows compatible WDM driver e g USB or FireWire camera MIL Video source accessed through the Matrox Image Library MIL e g analog or CameraLink cameras Canon Powershot digital still image cameras Canon Powershot series Canon EOS digital still image cameras Canon EOS Digital SLR series Prosilica GigE Prosilica Gigabit Ethernet cameras To initialize the video system choose either Video Initialize in the menu or click on the toolbar After initialization the live video will be displayed in the camera window You can switch between the live mode Live continuously grabbing video frames and the snapshot mode Snapshot freeze video in the Video menu Toolbar d Continuous image grabbing live video il Snapshot freeze video Notes Always select the live video mode to focus the image or adjust illumination If you prefer to work using the medeaLAB database see chapter Data management open the database before you initialize the video source This way the video view will be displayed within the main database form 10 meldea LAB IMAGING SYSTEM 3 2 Video options Before working with the system for the first time or if you have changed the in
4. Form Color Green s Included features JE Available features Color Blue Color Red Length Roundness width Click over the object to learn IMAGING SYSTEM Select the features you want to learn i e modify its Statistical descriptors Min Max Avg Dev in the Features list and switch to the Learn Mode by clicking the button Learn The dialog should now appear like in the picture on the left and a message appears in the Status line indicating the Learn Mode Now click on objects in the image which belong to the Germinated class and watch the statistical descriptors change You will not have to select all of the objects but it is a good idea to include objects representing the whole variety of the Germinated seeds class e g with short and long primary root After teaching enough objects leave the Learn Mode by clicking Learn again and the Status message will disappear You may set all descriptor values of the current class to 0 0 using the Reset button if you have to teach this class again from the beginning If you execute the menu command Image Search objects again now you will notice the new column Class in the result table and some of the objects should be classified as Germinated objects not fitting into any of your classes are displayed as NC Not Classified 32 SALES EEEEELELESLESLESLELEEESLESLESLEEESEEESESESESEEESE SE LREELEEEEESEEESESERESREERERESERE EE
5. SYSTEM All measurements may be carried out repeatedly based on a time schedule or external triggers in order to monitor dynamic changes of e g object area growth rate or colour decay Features for teachable classification may be choosen from a wide range of object parameters All measured values and classification results are displayed in spreadsheets and may be exported to Microsoft Excel automatically The integrated database allows easy management of result data and custom report generation Our imaging system also offers many interfaces for input and output Recording data from external sensors via AD DA converters data exchange with laboratory management systems OPC ODBC TCP IP or integration with PLC systems Beckhoff Automation components Siemens Simatic meldea LAB IMAGING SYSTEM 2 Installation Minimum system requirements are Personal Computer with Pentium 4 or better CPU Main memory 256 MB Free disk space 200 MB Operating system Microsoft Windows 2000 Windows XP or Windows Vista 2 1 Hardware and driver installation The steps of hardware installation will depend on your system configuration If you use an analog video camera it will be necessary to mount a Matrox framegrabber board to a PCI e slot in your computer In case of FireWire or USB cameras the standard PC ports are used and only driver installation may be necessary For digital still cameras using Stan
6. days button Continue unregistered or register your copy of medeaLAB Count amp Classify choosing the Register button IMAGING SYSTEM medeal AB is not registered This copy of medeaLAB is not registered Evaluation period expires in 10 days Press Register if you want to use medeaLAB after this period Continue unregistered For online registration you will need a valid email address and access to the internet Follow the four steps in the Register medeaLAB dialog Register medeal AB 1 Write down this ID number for copy it to the clipboard ID number ba 306931b563c8b3210f340f70ffade 2 Yisit the following address on the web http ww medealab de register english and input your ID number and email address on the web form 3 You will get an automatic response email with your registration code 4 Input the registration code from the response email and your email address here and press OK Registration code Email address same as in response email Cancel It is more convenient to use the computer where medeaLAB is installed for registration via internet and email because you may easily copy and paste the ID number WWW URL and registration code but it is also possible to transfer the data to another computer e g stored in a text file and the registration information you receive back to the medeaLAB computer Important Registration data are only valid for the cur
7. depends on your image and object properties which method yields best results The first method is contour based i e first object borders will be detected similar to the method in chapter 7 1 and then the maximum object extend within these borders will be determined see blue lines in the image below Activate only the checkbox for Length in the Options Formparameter dialog on the page Object not Width parameters Area and Formfactor are optional Make sure that no other object parameters are selected on the dialog pages and the Object Model Contour is selected on the 3rd page of the Formparameter dialog Parameter amp Model Then start a measurement using the menu command Image Search Objects Object axes willbe marked by a blue line and crosses and length will be listed in the result table 28 IMAGING SYSTEM SEES LESE EZELE ELLE LESE EEE EEE EEE EEE VEE l ey RRRRARRRARRASEEEEEEEEEEEEEREERESEEEEEREEEEEEEEEEEEEE MRR RRR RRR RRR RRR RRR RRR RRR R R T E Count Result 5 odes os ipa Aa Tm Form eng ves Border LEAT SAAR EEE TEEN ET ET ET REESE ES EES SEE EEEEEEEEEEEEEEEES LER 1 1060 244 3195 00 4 60 500 86 True 2 3518 00 456 15 s 3 991 576 2955 00 21 41 230 70 22 True i 1054 694 5499 00 26 67 441 41 19 True 5 949 775 1535 00 4 96 101 83 7 True Sum 16708 11 1 6e 003 Root length measurement using the contour based method with additional object
8. measurement data In the Open New File dialog you may select an existing file which will be overwritten or type a new filename 9 1 Measurement Options for image analysis On the Measurement Options tab please select the measurement method source of data values and time parameters Configuration Measurement Options Signal Control Configuration Method Y Axis Chan1 AOl A Chan2 AOl1 Chan3 AOl1 Chan4 A0l1 Measurement Timing Interval 0 133 min sec msec Run Time 0 hour min sec For the X axis of the resulting chart you may choose between time scale and measurement count medeaLAB offers you the following image analysis measurement methods to retrieve data values for the Y Axis 34 meldea LAB IMAGING SYSTEM 9 1 1 AreaChannel The area of the largest object in the image from the selected video channel on multiple camera systems is measured The area is calculated from the object contour If you have defined multiple Regions Of Interest ROIs only the first one will be used If this method is selected you may also specify one or more video channels in the list below Chan channel 9 1 2 AreaAOI As with the method above the area of the largest object is measured but in one or more ROIs selected in the list below AOI1 AO12 9 1 3 MeanColor The mean color gray values of the ROIs selected in the list below are measured 9 1 4Distribution The mass distribution of obj
9. the Windows XP directory C Documents and Settings All Users Application Data Medea AV MedeaLAB script or the Windows Vista directory C ProgramData Medea AV MedeaLAB script Al v fe Index E Count Result 3 Ind Length Wid Ti Nod Bord 2 a r i ndex Leng salaa SS EBENEN Index Length Width Tips Nodes Border 1 258 00 3 00 3 1 True 2 11900 300 5 2 7 1258 3 3 1 WAHR 3 15506 4002 0 Tu 3 2 119 3 5 2 WAHR res an 4 3 155 3 2 0 WAHR Sum 532 9 10 3 G Result table data in medeaLAB Same result table data after execution of the File Export Excel command 10 2 Database management For data management the Microsoft Access database medealab mdb is installed together with the program All single measurement results are stored into this database for later analysis reporting or export Measurement results are grouped into measurement series You may also assign additional information to each measurement series Open the Database window with the menu command File Open Database or the leftmost icon on the database toolbar M lt aX amp sa The database toolbar offers buttons for navigating through the database adding or deleting records and printing reports Description of the buttons from left to right Open database show first record show previous record show next record show last record create new record delete current record count manual by mouseclicks show report 2
10. to modify existing ROIS you may choose the menu command Image ROIs Modify which opens the Region of Interest ROI Control dialog E DSCN2438 JPG OLE E IEEE EEE EEE EEE EEE Region of Interest ROI Control fx Position x 367 y jes Size width 1348 height f1350 Ange fo Width fpo Raster x jo y lo Border fo Color Color Change Position ym SelectActive Select All Modify Add off MEABRRRRRRRRRRRREREEEEEELEELELEEELEEEEREEEEEEEEEELS BEEREEEEERELEELELEEEEEEEEEEEEESEEEEEEEELESEEEEY r A A A A A A A A A A A A A A A A A A A A A er Bu A A A A A A A er er er 2 i A f A 7 A f A A en Al ra A A A A A A er A A A A A A A A A A A A A A A A z f EE ee Image with two ROIs 1 Ellipse 2 Parallelogram and the Region of Interest Control dialog Elements of the Region of Interest ROI Control dialog Position x y and Size width height In these fields the position relative to the top left image corner and the size of the active ROI is displayed in pixel and may be modified button Modify Angle Angle of orientation only elliptic ROI Width Width of ring only Ring ROI Raster Rectangular ROIs may be devided into columns x and rows y of grid cells This is useful for evaluation of object areas in each cell e g plant trays Border 20 IMAGING SYSTEM
11. 972 IMAGING SYSTEM A window with the current measurement data and buttons to execute measurements is displayed to the left of the image window after opening the database Measurement Row pozersihed Measurement Wasser N Information on the current measurement like name of Date Time 21 01 2010 12 17 12 associated measurement row and used parameter set Day 1 profile Fill Level 100 mil Image 757 jpg Profile BWIN eur Results ee Count number of objects computed area and Area Total p 0 000 classification values Area Sum p 261464 000 Area Avg um 140732 000 Aree Dev pm 199023 680 Area Median 281463 000 NC 100 Count button Creates a new database record executes measurement and stores results in the new record Repeat button Executes measurement and stores results in the current raises EN record Save Image checkbox If checked the image used for measurement will be stored as a JPG file in the images subdirectory for later analysis or documentation Measurement number Index of the current measurement and number of measurements in the current measurement row J Save Image Measurement number The You may add change and switch between measurement rows using the menu command Database Measurement Rows Each time you execute a measurement using the Count button you will be prompted for a descriptional text for this measurement Measuremen
12. Bare CL IFAN IMAGING SYSTEM medeaLAB Count amp Classi Image Analysis System User Manual Medeal Count amp Classify 704 Fie Options Macro view Window Help Sala e alu e Z als ofo alz r km 704 bmp 87 83 normal 87 83 normal 88 84 normal 8g 81 normal 92 85 normal 83 normal normal normal normal normal normal normal normal normal normal normal normal normal normal normal normal normal normal normal normal hellgrun hellgrun normal normal normal normal normal normal normal normal z5 x 193 y 11 RGB 198 251 254 Language Version English meldea LAB IMAGING SYSTEM Medea AV GmbH Am Weichselgarten 23 D 91058 Erlangen Germany www medealab de software Version 6 6 Date February 2010 meldea LAB IMAGING SYSTEM Copyrights and Trademarks Medea AV Multimedia und Software GmbH 1998 2010 Medea AV GmbH All rights reserved No part of this publication may be reproduced or transmitted in any form or for any purpose without the express permission of Medea AV GmbH The information contained herein may be changed without prior notice Microsoft Windows Excel and Word are registered trademarks of Microsoft Corporation HTML XML XHTML and W3C are trademarks or registered trademarks of W3C World Wide Web Consortium Massachusetts Institute of Technology All other product and service names mentioned are the trademarks of their respecti
13. G SYSTEM 1 Introduction medeaLAB Count amp Classify is an image analysis system for color or grayscale images which is able to detect count measure and classify objects Due to its flexibility it is used for a wide range of applications in industry and research Our website http www medealab de will give you an overview of medeaLAB applications medeaLAB Count amp Classify works with all kinds of image sources Image files in various formats analogue and digital video sources video cameras VCR FireWire USB or Ethernet cameras still image cameras remote control of Canon and Olympus cameras scanners TWAIN interface or video files medeaLAB Count amp Classify may be combined with macroscopic and microscopic optics camera mounted on microscope due to its calibration capabilities even perspective transformation Processed images may be stored in various formats for documentation purposes BMP TIFF GIF PNG JPG JPEG2000 etc Configuration of thresholds and search parameters is easily done using interactive and automatic methods object detection is performed automatically For each object many parameters are evaluated area form factor roundness mean gray color value and variance center of gravity main axes convexity orientation length of appendices etc For image enhancement medeaLAB offers many user configurable filter functions like smoothing Laplace Median Edge filters etc IMAGING
14. LEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE EEE Bi SN SNS ASA SAA SA AAAS SSA SANS AA AAA AANA SS SW WW WW AW hh A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A 4 SS SS SS eee ee eee A es FR Se nn En a a a ne FALL LLLES ET EZELLL ZELL LT LLC LLC LT LE LT LT LLC LLC LE LT LT LT LE AL LE LE LT LE LT LT LE LS LE LT LT LH EEE LE LT LT LH LS LLC LT LT LT LT LLC LH LS LE LT LH LS LH OEE EEE Image with some of the seeds already germinated First you have to decide which object features are characteristic for germinated seeds and select them for evaluation We clearly see that germinated seeds differ from other objects detritus and ungerminated seeds in area they have grown form irregular shapes color greenish appendices 30 Formparameter Object Nucleus Parameter amp Model Object Parameter Parameter Min Action Max Action mE Formhactor 1 5 O Length 0 1 100 O width Dio 100 Color Avg O0 00 00 FF FF FF LI Homogenity 0 1 L Convesity D J DL ppendix z Exclude iv Hit Boundary Default Load Save Abbrechen Hilfe IMAGING SYSTEM Therefore in the Form Parameters dialog we check at least Area Formfactor and Color We do not have to set lower or upper feature limits Action but you should chec
15. RE EERE EERE REESE U U U EEE EEE U N N FE sei BR E Count Result 40 988 318 Germinated 2 934 327 Germinated 13 1068 325 NC Germinated objects are marked with a green border which was selected as the color for this class using the Color button right to the Classname in the Classes dialog Strategies for optimal classification results If germinated seeds are not classified correctly as Germinated but e g as NC add these objects to the Germinated class in Learn Mode If objects that do not belong to the Germinated class are misclassified as Germinated try the following 1 Assign more suitable objects to the Germinated class in order to improve the statistical descriptors for this class 2 Create and teach also classes for the objects not classified NC This will lower the risk of misclassification because an object is assigned to the class to which it fits best 33 IMAGING SYSTEM 9 Time series measurements To start a time series measurement use the menu command Measure New or the toolbar button ad First select a measurement method Select Method AD Measurement analog values from an A D converter as a data source AD Measurement Video measurement image analysis results as a data source Video Measurement Cancel Before you can start a time series measurement you will have to create a new file or open an existing one which will hold your
16. Tn Si i a a a a a a a Sa TT L W Count Result 17 Zi 0 0 5 ie ooo oo oo 0 Fe MMA ee ge Ta Te Ta Ta Ta Ta Ta Tn Ta Ta Ta Ta Tn Ta Ta Ta Ta Ta Ta T a Te a e Sr a a a oe oi oo a L FrFFIHFERFFFERFERFFRRRFERFRRR Example 2 Conglomerate cutting using Convex Defect Analysis on the same image with sensitivity 0 results in only 12 objects 25 IMAGING SYSTEM 7 Basic steps for counting and measuring objects Before executing measurements the following system settings chapter 4 should be adjusted System calibration Pixel calibration Gray level color thresholds You can only execute measurement functions of the Image menu if an image window is active click on the image window to activate it 7 1 Counting objects using contour analysis How to get an idea of how to set reasonable lower and upper form parameter limits for counting contour objects Activate only the checkboxes for Area and Formparameter in the Options Formparameter dialog on the page Object Make sure that no other object parameters are selected on the dialog pages Then start a measurement using the menu command Image Search Objects Object area and form will be listed in a table E Count Result 29 Index Posx_ Posy Area um Form Flagellum Moving 17 Count evaluation The system will mark contours of all objects within form parameter limits in red all oth
17. conglomerate cutting may be used as a preprocessing step before object measurements in order to increase precision of object count results You may choose between two different algorithms Convexity defect analysis Watershed analysis Select Options Object Filter Conglomerate Cut from the menu to display the Conglomerate cutting dialog 23 Conglomerat cutting 1 cs 5 T e e Te Se S Se Tae Ta Sa Ta Taa Taa as Taa Taa ee Se Of Convex Defect Analysis Watershed Sensitivity 1 Er i eee a ee oe Sa i i a rn Cancel seeeeecsessetsces 3 Ki m I s a a ina iis ia a Dis Die Dis Dis Di Dia Dia Di Dia Dia Pe a a es Dr Da a in 4 TALATI HH He He He He He He Me He Me He Me Te m He Me He Me Oe p FFERFEERSEEESSEESSERSSEREEEES A IMAGING SYSTEM Convex Defect Analysis searches for possible locations to split objects depending on convexity deviations of contours Watershed objects are split using the Watershed algorithm based on the topology of the image gradients Sensitivity Range is 0 10 10 means highest sensitivity Reduce sensitivity if objects are split into too many parts W Count Result 15 a fx _ OO 0006060600606 060 06 0 00 O m Example 1 Conglomerate cutting using Convex Defect Analysis with sensitivity 5 results in 15 objects 94 1 5 Pos E E MT Te Te Te a Ta Ta Ta Ta Tn Ta Ta a Tn m na ma Ta n
18. cript subfolder see chapter Maintenance 39 meldea LAB IMAGING SYSTEM medeaLAB is shipped with a sample XLS report template file Report_Template xls in the MedeaLAB Data subfolder see chapter Maintenance Please contact the medeaLAB support if you want to modify existing templates or create new ones 40 LAB i IMAGING SYSTEM 11 Maintenance It is strongly recommended to frequently save backup copies of the medeaLAB settings as well as of your data During installation of medeaLAB Count amp Classify the folder Documents and Settings All Users Application Data Medea AV MedeaLAB Windows XP or ProgramData Medea AV Medealab Windows Vista is created on the installation drive to store the settings of program modules and your data Where medeaLAB stores information by default Windows XP Documents and Settings All Program settings registration information Users Application Data Medea AV Medealab and profiles INI XML AOI and CLASSES Documents and Settings All Database MDB file report templates XLS Users Application Data Medea and HTML files AV Medealab data Documents and Settings All Stored image files Users Application Data Medea AV Medealab images Documents and Settings All Script files for automating tasks Users Application Data Medea AV Medealab script Windows Vista and later ProgramData Medea AV Medealab Program settings regi
19. cs rather than the chamber depth in the depth field of the calibration dialog 4 2 Setting of the gray level or color intensity thresholds There are two ways to adjust the gray level or color threshold values for object segmentation in the medeaLAB system Automatic adjustment Manual adjustment using preview In most cases you will only need the automatic method It is also a good starting point if you want to do fine tuning using the manual method 4 2 1 Automatic adjustment 1 Make sure that you have selected live video mode your investigation objects are in focus and well illuminated 2 Take a typical snapshot of your objects from the live video or load an image from file or TWAIN source 3 Choose the Options Thresholds menu command and indicate whether the Object Appearance is Bright against a dark background Dark against a bright background Middle against a dark and bright background or Dark Bright background is in the middle The correct setting is important because it is the only clue for the system which objects it should search for Thresholds Channel 1 Color Thresholds lower level high level Red gt 20 al u gt 149 Green 2 4 08 Blue lt yas E S Automatic Offset 120 gt lt lt gt Object Appearance C Dark Bright Middle Dark Bright Default Learn Object Learn Back Preview Cancel Expert Ready 4 Click the Automatic button to set t
20. dard Windows interfaces no driver installation is necessary Please refer to specific Installation Notes delivered with your system In most cases the system will be delivered ready to use and no hardware configuration is necessary unless the whole system must be reinstalled Please contact the medeaLAB Support Team before you try to reinstall the system on a PC Important information regarding all types of cameras No matter if the camera is switched on or off never aim at the sun or other extremely bright objects Otherwise blooming or smear may be caused Clean the CCD faceplate and or lens very carefully Do not clean the CCD or lens with strong or abrasive detergents Use dry air lens tissue or a cotton tipped applicator and ethanol 2 2 Program installation Important To install the software administrator rights are essential To install the medeaLAB Software put the installation CD into your drive The medeaLAB setup starts automatically otherwise start the installation by double clicking on the file setup msi in the CD root directory Choose Typical setup and follow the instructions Before you can use the medeaLAB software you should also install all necessary drivers see 2 1 After installation you may have to restart your computer 2 3 Registration When starting medeaLAB for the first time the message medeaLAB is not registered appears You may continue to work without registration for a period of 30
21. e E ee 30 9 TIME series measurements ussussussseesennennnnnennnnnnnenennennennnnenennennennnnnnnenennennennnnenennennennnnennenennennennnnen 34 9 1 Measurement Options for image analysis ccccccee eee ee eee eee ee eeeeeeeeeeeeeeeeeeenegs 34 9 1 1 PCAC Fe EEE 35 APEE 2 E pe ee TO E SE 35 TES Medne O Ghent enn A A EEA ee ee ee ee ees nn sees ne nee ee 35 ITa RSTO accra teers nennen EEE 35 Il RU A ee EEE 35 FTO ON EEE N se ennesnnesce panes anesasansesnnencsnnscancasscane 35 9 2 SIG CO On ea nee ne ea een an ren 35 9 3 COME GU AU OT ae ee een 35 9 4 Start the measurement nee 36 10 Data management ussuenenensenenennnnenennnnnnenenennn nenn nnnennn nenne nen nnnnenenenennenenenenennnnenensnennnnenennnennenenennn 37 10 1 Data GX DOUG ee rer ae su seen E ee ers scaucens st 37 10 121 EPON O FNC ee ee 37 10 122 EX DOR TO WVINCIO SOI ERCE ee 3 10 2 Database management icccccccccccccccceceeeeeeeeeeereeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeees 37 10 32 WRCDONRCS oer ietcaene sees csecebrasteesoeaessae tee 39 WOES TIVE FS DOM ee 39 WOES ALS TS cede crcnceeeseoe see sessions iadeea den cdnecdeeccecacededececesedacedsteeaaneesecacececesacesecedensceandecsteecdeactecace 39 kl MamtenaN Eissii nada ncaa aadee anne nmeceeacaneeanaaatecee ace tacameaceeceratcdonsteeas 41 12 Communication with external devices cccccccccsecccseeceeseeceeeeeueeseeseeeseeeseueesueesaeesaeeteueeeueeaeeesaeess 42 IMAGIN
22. ects in the image is measured The center of gravity of each object is calculated and their distribution is analyzed 9 1 5 Curvature The maximum curvature of the largest object in the selected ROI is measured The curvature will be determined from tangents approximated to the object contour calculated from n 2 contour points on each side of the current contour position where n is the total fit width Curvature measurement with markers on object end points and point of maximum curvature 9 1 6Growth The translation of the center of gravity of the largest object between subsequent images is measured The translation value is the euclidean distance between the center point at measurement start and its current position If a ROI is defined it will be moved after each measurement so that the ROI center matches the object center 9 2 Signal Control On the Signal Control tab you may find advanced options for triggering external devices during measurements 9 3 Configuration On the Configuration tab you may choose options for image handling 35 IMAGING SYSTEM Configuration 53 Create AVI File save images of all measurements to digital video file AVI Measurement Options Signal Control Configuration format Fl Create AVI File Save BMP File save each E Save BMP File measurement image to bitmap file I Activate Filter Blur Median BMP format Activate Filter smooth each image before measureme
23. ers in green In the title bar of the table window the number of objects counted is displayed By clicking on a table row the corresponding object in the image will be surrounded by a box Vice versa when clicking on an object in the image the corresponding table row will be highlighted By this get an impression of appropriate maximum and minimum area and formfactor parameter limits With some tolerance use the area and formfactor values occuring in the measurement to adjust the Min and Max columns of the Formparameter dialog according to your needs 26 IMAGING SYSTEM If unwanted objects are listed in the table e g dirt or detritus you may have to narrow the limits of your filter settings If some objects of interest are not listed you may have to widen the limits of the filter settings Fine adjustment Even if thresholds and filters are set appropriately it might occur that individual objects are excluded from the analysis thus marked in green whereas occasionally an unwanted object will be included thus marked in red In these cases click on one of the objects with your left mouse button the object will be marked by a box then click the right mouse button and select Include Object or Exclude Object from the menu Based on your changes the system will automatically adjust parameter limits If you want to separate adjacent objects use the Conglomerate Cutting functions chapter 6 3 7 2
24. eseeeceeecsueeeueeseeseeessueeeueeseeeseeesaass 12 4 1 PIXEI CE DIA EB OT een en EP a nn ee ee ete 12 4 2 Setting of the gray level or color intensity thresholds ssss2s00000000nnnnnn nun 00 13 4 2 1 AUTO AUG GIGI US UCI een ee 13 422 Manlaladjusimen sneeenn ee 14 4 3 Using object features for identification and discrimination zssssssnnnnnnnn none nenn 15 9 SBEIECLON ON MAGEN EE E E tee eee erate ea eee seed ee ee eee 19 6 Methods for preprocessing image Aata cccccccecccsceceecceecceeeceucceeeeceeecuecsueeseeeceeeseuessueeeseeseeeseuesseeenes 22 6 1 DIGIC al 1 E Sasesenssancantastneadwanancad E A E T E E E URL ETERR 22 6 2 Mage M Ora O eose E E E E E E 23 6 3 Separation of object aggregations conglomer tes ccsseeeeeeeeeeeeeeeeeeeeeeeeeeeegges 23 7 Basic steps for COUNTING and measuring objects ccccseccceeeceeeeeeeeceeeseueceeeeseeeseeseueeeueeseeeseeeseueeeeeeaes 26 7 1 Counting objects using contour analySiS uuuuuuuunnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 26 7 2 Counting coherent objects using blob analysis ccc cece cece eeeeeeeeeeeeeeeeeeaeaaaas 27 7 3 Measuring area fractions within ROIS cccccccce cece cece eee eeeeeeeeeeeeeeeeeeeeeeeeeneenegs 27 7 4 Length measurement s ssesssssenrsnrrnrrnrsnrsnrnnrunsunrunrunrunrunrunrunrunrnnrunrunrunennrun 28 14 Manual OBJECE COUNTING nennen een een 29 8 9 2 6 CIASSIl IC AO 9 Pee ee een eee en te ee n
25. esholds are parameters for the Lines filter line detection using the canny algorithm Note If the size of the kernel increases the filter effect e g Soften smoothing increases But if the neighborhood considered is too large blurring and other unwanted effects might appear in the image In case of smoothing low pass filtering the selection of kernel width is a compromise between reduction of noise and a low blurring effect The last filter operation may be cancelled by the menu command Edit Undo DD IMAGING SYSTEM 6 2 Image integration If your image source is a video from file or camera and you have to work with low light intensities e g fluorescence or x ray imaging you will often face the problem that your objects have low contrast or the images have a lot of background noise In this case you may use the function Image Integration which computes the sum or average of an image series Select Video Image Integration from the menu to display the Image Integration dialog Image Integration Number of Images Method Average f Addition at With the above settings the pixel intensities of next 3 video frames will be sumed up and then image capture will stop showing the result image for further processing X ray image of animal scull Image enhancement after integration addition of 3 video frames 6 3 Separation of object aggregations conglomerates Separation of adjacent objects by
26. filter area gt 1000 The second method is skeleton based Select Skeleton as Object Model on the 3rd page of the Formparameter dialog Parameter amp Model Then start a measurement using the menu command Image Search Objects Object axes will be marked by a red line and length will be listed in the result table CPI III ILI IEEE EEE EEE EEO PPP TETT Ver a ay i Bis lt s A o sy A A A 2 ba 4 7 f yo Bet AS es AOE OE EEE ETAT EE EE EEE EE EE EEE EEEE I4OLES EL F of Aue Root length measurement using the skeleton based method 7 5 Manual object counting This is a helper function in order to avoid wrong count results After selecting the menu command Image Count Manual you may click on single objects The objects are marked with a cross and assigned a number and a counter will be incremented You may also undo a single object selection with the Remove button 29 IMAGING SYSTEM 8 Object classification medeaLAB offers features for flexible object classification First you teach the system which objects belong to which user defined class During this selection process a statistical classification scheme is computed which will be used in subsequent automatic classifications We will demonstrate these abstract processes as follows Example Classifying seeds as germinated PMLCL EEE EE EE EEE EEE EEE EEEEEEEEEEEEEEEEEEEEEEEEEEESESEEE MELEEEEEEEE
27. ght position and release the mouse button there create a parallelogram ROI Press the left mouse button at the first corner position in the image window hold it move the mouse cursor to the second corner and release the mouse button there Then move the mouse cursor to the third corner position and press the left mouse button The fourth corner is automatically determined from the position of the other three corners and the ROI is drawn only if all corners are inside the create a freehand ROI Either draw the border of the ROI with the left mouse button held down or select several edge points of the ROI by clicking the left mouse button Clicking the right mouse button afterwards will close the ROI polygon create a ring ROI Same as Ellipse the width of the ring is adjustable in the Region of Interest ROI Control dialog menu Image ROIs select a particular ROT Click with the left mouse button in it An active ROI appears highlighted by a hatched border Drag the small black squares on the ROI border with the mouse only ROIs of type 19 IMAGING SYSTEM po rectangle or ellipse move ROI Click with the left mouse button inside hold the button and move the ROI to the new within the image delete ROI Select the ROI then choose the menu command Image ROIs Delete or press the associated toolbar button or press DEL on the keyboard In order to create multiple ROIs of exactly the same size or
28. he thresholds automatically with the parameter Offset the size of objects is increased after automatic adjustment because the automatic method normally leaves the objects too small 5 Click the Preview button to check the binarized image in the camera window Your investigation objects should appear black on white background in preview mode s49 meldea LAB IMAGING SYSTEM Preview mode is ended by clicking the Preview button again Notes If you use a video image source you may also watch the video images being binarized continuously Just click the Pseudo Live button in Preview mode The automatic adjustment is done using the pixels in the current ROI region of interest area or the whole image if no ROI is set For more information about regions of interest see Selection of image regions chapter 5 below 4 2 2 Manual adjustment 1 Make sure that you have selected live video mode your investigation objects are in focus and well illuminated 2 Take a typical snapshot of your objects from the live video or load an image from file or TWAIN source 3 Choose the Options Thresholds menu command and indicate whether the Object Appearance is Bright against a dark background Dark against a bright background Middle against a dark and bright background or Dark Bright background is in the middle 4 Click the Preview button to check the binarized image in the camera window Your investigation objects sho
29. ing an HTML report template a report window is displayed in medeaLAB At the bottom of this window you will see this button row Copy to cipboard Print shows the Print Options dialog and allows printouts Save saves the report in HTML MHT web archive or TXT unformatted text format Copy to clipboard copies the whole document to the Windows clipboard you may then paste it directly into e g Microsoft Word Close will close the report window A right mouse click in the report window will give you additional menu options e g print preview A medeaLAB HTML report template file consists of normal HTML code with placeholder keys enclosed in into which medeaLAB fills in report data medeaLAB is shipped with two sample HTML report template files count html and lemna html in the MedeaLAB Data subfolder see chapter Maintenance Please contact the medeaLAB support if you want to modify existing templates or create new ones 10 3 2 XLS reports If you execute the menu command Database Print Report after selecting a XLS report template Microsoft Excel will be started automatically and the report data will be inserted into the template A medeaLAB XLS report template file is a normal Excel spreadsheet The name of the current measurement row will be inserted into cell B2 the measurement data in the cells starting in row 4 This behavior is controlled by the script file Report txt located in the MedeaLAB S
30. jects are not counted if their contour hits the border of the image or ROI region of interest All object feature filters may be separately activated or deactivated using the checkbox in the column Parameter Edit the lower filter limit in the column Min the upper filter limit in the column Max by clicking the left mouse button within the field Both limits may be separately activated or deactivated using the checkbox in the Action column to the right of the Min Max column The Action checkboxes are only visible in the current row parameter Width in the picture above in all other rows is displayed if the filter limit is active and if not So with the settings on the above example screen only objects which meet the following conditions will be counted Area between 100 and 500 square units formfactor above or equal 1 5 maximum length and width of 100 units all other filters are switched off If you also want to evaluate the inner structure of objects you may activate the following filters on the 2nd page Nucleus of the Formparameter dialog 16 IMAGING SYSTEM Formparameter Object parameters Object Nucleus Parameter amp Model Number of Nuclei Number of nuclei Object Parameter distinct regions inside the object Number of Nuclei Cell Nuclei Area Ratio Ratio of cell area ee uss Ama lato to the area of the nuclei without unit Area er Area Area of nuc
31. k Exclude Hit Boundary because it is not possible to classify partial objects Now you should execute a test for object detection using the menu command Image Search objects If the objects desired are not marked correctly and listed in the table please check for the right Thresholds and retry The next step is to teach the system which objects belong to a class Activate the image window and open the Classes dialog using Options Class definition on the menu Classname xl Color Au Moai ese Features Avg lt a Included features Available features Area Color Blue Color Green Color Red Form Length Roundness width At the first time no class is defined so type a name for your new class in the Classname field and click Add As a default the Area feature is added automatically to the Included features list for the new class You now may add features characteristic for your objects from the Available features list with the button or remove from the Included features list with the gt button In our example we add Form and Color Green besides Area At least these features should have been selected for evaluation in the Form Parameters dialog before You are free to select more features for evaluation in the result table but only the mentioned three features will be used for classification 31 Classname Germinated ka cat Delete Avg Features Area 2
32. keeping the left mouse button pressed 4 When you release the mouse button the Calibration dialog will appear Calibration Reference Yolume 93 33333 width um 94 4444400 Height pm Chamber Depth pm Yolure mi Conc Number 4 72 Mio Save OF Cancel 5 Use the Calibration dialog to transform the distance in pixels to a metric unit Choose the relevant metric unit um mm cm Type the length of the line you have drawn e g 100 um into the right field below Ratio For automatic calculation of concentrations e g cells ml you also have to set the depth of the chamber or the fluid level Type the depth e g 10 um into the field depth of the calibration dialog 6 Check size and area values of your investigation objects for plausibility measured by e g Image Search Objects after all adjustments in this chapter are done If you get wrong values please try to repeat all steps of the pixel calibration Note To achieve good concentration results the field of depth of your optical system range in which objects are in focus should correspond to the chamber depth If you use a chamber that is deeper than the field of depth of the optics objects outside that field will not be visible clearly and therefore may be not counted by the medeaLAB system This would result in concentration values being to low You may compensate this effect by entering the field of _ 2 IMAGING SYSTEM depth of the opti
33. leus in calibrated units Formfact Formfactor Roundness of nucleus a ir eS without unit u Erz Color Avg Mean color value of all Homogenity nucleus pixels Homogenity Uniform distribution of color gray values inside nucleus calculated from variance values without unit On the 3rd page of the Formparameter dialog Parameter amp Model options for contour analysis and five different object models are available Object parameters armparameler amp Holes Subtract holes background Object Nucleus Parameter amp Model inclusions within objects from object area Object Parameter Isthmus Points Identify constrictions and V Holes mark them Convex Analysis Search object contours M Isthmus Points for convex segments and mark them Convex Analysis Concave Analysis Search object contours Concay Analysis for concave segments and mark them Split Contour Split Contour Split contours into convex moos D Maximum Distance and concave segments thus creating new objects with results like aperture angle endpoint distance etc Cluster Near objects are regarded as one Merge Contour a cluster object depending on Maximum Kontour Distance Edge Object Merge Contour Objects cut apart by C Ellipse raster cell borders of raster ROIs will be ol merged if they meet several conditions Object Model Choose the method for segmentation which is appropriate to evaluate
34. nt 9 4 Start the measurement Start the measurement using the menu command Measure Start 1800 Light 1600 1400 1200 1000 800 600 400 200 1 0 2 0 3 0 4 0 5 0 6 0 3 273 T me day hour Chart of brightness data measured over a time period of 7 days and 13 hours the time axis is scaled automatically You may interrupt a time series measurement at each point in time by choosing the menu command Measure Stop 36 meldea LAB IMAGING SYSTEM 10 Data management 10 1 Data export medeaLAB Count amp Classify offers two file formats for table data export and direct data transfer to Microsoft Excel 10 1 1 Export to file Choose the menu command File Export ASCII File to write the current table data to a plain text file Choose the menu command File Export CSV File to write the current table data to a comma separated value CSV text file with as a delimiter between values 10 1 2 Export to Microsoft Excel Choose the menu command File Export Excel to transfer the current table data to a Microsoft Excel spreadsheet This transfer method does not need a file for data storage but Microsoft Excel is started automatically and a new spreadsheet with the medeaLAB data is created Therefore Microsoft Excel must be installed on your system The transfer is controlled via the export script export txt this file should be in your script directory usually located in
35. rent Windows installation and the hard disk on which medeaLAB is installed You may use the same registration code only for reinstalling medeaLAB under the same Windows installation at the same location drive as before Registration information is stored together with all program settings Therefore it is strongly recommended to save a backup copy of all program settings see chapter Maintenance 2 4 Basic software configuration IMAGING SYSTEM Before working with the program you may select paths for storing image files and exported data see chapter Maintenance for default locations After the program has started choose the menu command System Options In the System Settings dialog edit the Images and Data paths System Settings Images Iz Release image Data ZAR eleaze data E Frotiles Load parameters BIWIN 5 ave Delete IMAGING SYSTEM 3 Video and image settings To load an image into medeaLAB Count amp Classify you may initialize a live video source and take a snapshot open an image file from disk paste an image from the clipboard menu option Edit Paste As New Image or capture an image via the TWAIN interface menu command File Aquire from a connected digital camera or scanner selected by menu command File Select Source 3 1 Initialize live video First you have to select a video source in the list on the toolbar Video File Video
36. stration information and profiles INI XML AOI and CLASSES files ProgramData Medea AV Medealab data Database MDB file report templates XLS and HTML files ProgramData Medea AV Medealab images Stored image files ProgramData Medea AV Medealab script Script files for automating tasks You may easily backup all medeaLAB settings and result data by making backup copies of the whole MedeaLAB folder and its subfolders 41 meldea LAB IMAGING SYSTEM 12 Communication with external devices In the Utilities menu you may find options for communication with external devices medeaLAB Count amp Classify is able to import data from several types of A D converters menu option Utilities Select Hardware The external sensor connection is configured via the Utilities Sensor Setup dialog medeaLAB Count amp Classify is also able to communicate with the medeaLAB precisionScan and medeaLAB trayScan devices See www medealab de for more information on this devices The medeaLAB precisionsScan illumination and motor driven facilities are controlled via the Utilities Scan Chamber dialog only available if driver is installed and device is connected Backligh fp E Save ave Delete ES Advanced Settings explanation in the precisionScan User Manual shipped together with the device 42
37. t Ex Description max 255 Characters cence _ You may skip editing the description by simply pressing the lt RETURN gt key or clicking the OK button clicking Cancel will abort the measurement All measurements of a measurement row may be displayed in a table view menu command Database Show Table or as a formatted report especially for printouts menu command Database Print Report 38 meldea LAB IMAGING SYSTEM 10 3 Reports A report with the measurement data of the current measurement row will be generated if you select the menu command Database Print Report E HH or the button in the toolbar medeaLAB supports two different mechanisms of report generation Reports assembled from HTML templates htm html containing placeholders for result data This method is most suitable for generating reports with formatted text and logo images Reports generated by loading a template Excel file xls remotely to Microsoft Excel and transferring result data directly to the spreadsheet This method is best Suited if you want to analyze your results statistically afterwards For both methods you will have to select a template file first menu command Database Select Report Template otherwise the Print Report command will lead to the message Error in report 10 3 1 HTML reports If you select the menu command Database Print Report after select
38. uld appear black on white background in preview mode Clicking the button Pseudo Live in the preview mode shows the binary image in a live simulation available only for video input sources 5 If necessary adapt the thresholds by using the Ea buttons for shrinking or increasing the object area in preview mode These buttons work on all three color channels and change the threshold in steps of 5 units 6 You may also do fine tuning in steps of 1 unit with the arrow buttons to the left and right of each color channel Example 1 If Object Apparearance is set to Dark you can adjust only the lower levels and the high level sliders are diabled see left picture below because only the lower thresholds are used Set the lower level to 20 for all channels Red Green Blue Result The system will count all pixels with values greater or equal O and less or equal 20 for all channels as object pixels Example 2 If Object Apparearance is set to Bright you can adjust only the high levels and the lower level sliders are diabled see right picture below because only the high thresholds are used Set the high level to 149 for all channels Red Green Blue 14 IMAGING SYSTEM Result The system will count all pixels with values greater or equal 149 and less or equal 255 for all channels maximum value as object pixels Thresholds Channel 1 Thresholds Channel 1 Color Thresholds Color Thresholds lower level high level lower le
39. ve companies Data contained in this document serves informational purposes only National product specifications may vary These materials are subject to change without notice These materials are provided by Medea AV GmbH for informational purposes only without representation or warranty of any kind and Medea AV GmbH shall not be liable for errors or omissions with respect to the materials The only warranties for Medea AV GmbH products and services are those that are set forth in the express warranty statements accompanying such products and services if any Nothing herein should be construed as constituting an additional warranty sd LAB i IMAGING SYSTEM Contents te OOU CO aee E E R E R E 5 Be 6 5 21 21 6 9 ee ee nee eee ee nee te tne eee eee nee ee 7 2 1 Biclechcleceslaemolais allgceailicis 6 ae en ee a ee en eT 7 2 2 Programm INSEANOC ON saiciacagadtancacancaan EEETEEURFENEETUFSLOUSEUFFUHEEREERLOUSEUFELHEEFEERERERE 7 243 ReQISTLALION PATEE E ea ee ee a a aa a re ent 7 2 4 Basic software configuration ccccccc cece cece eee eee nennen nennen nennen nennen nennen nennen nennen 8 3 Video and image settingS 2 u024020000000nn nun nenn ann nun anne nnnnnnn anne nnnnnnnnnnnnnnnnnnnnnennnnnnnnnnennnnnnunnnennnnnnnnnn 10 3 1 Inlielalize VO VOCO ee eee ee ea E ea aaa 10 3 2 VIGGO ODEON a vies a a I ET EI ER ae 11 4 Setting of system parameters ccceccseccceecceeeceeeceeeceueecueecseccuecsueesue
40. vel high level Red gt 4 gt 149 Red lt wim a _ gt 149 Green af gt 20 KE 149 Green 4 20 al gt 149 Blue Ya gt 149 Blue lt Ya 2 4s Automatic Offset 120 gt lt lt gt Automatic Offset 20 gt lt lt gt Object Appearance Object Appearance f Dark Bright Middle Dark Bright C Dark Bright Middle Dark Bright Default Learn Obiect Learn Back Preview Default Learn Obiect Learn Back Preview Cancel Expert Cancel Expert Ready Ready Object Appearance Dark Object Appearance Bright Additional options Default The Default button sets all lowel levels to 80 and all high levels to 240 Learn Object If the Learn Object mode is activated you may adjust the thresholds automatically by moving the mouse cursor over objects with the left mouse button pressed It is important not to touch background pixels during this mode Learn Back If the Learn Back mode is activated you may adjust the thresholds automatically by moving the mouse cursor over the backbround with the left mouse button pressed It is important not to touch object pixels during this mode Expert Access to the Threshold Expert Settings dialog where you may store thresholds to a text file or load thresholds from a text file 4 3 Using object features for identification and discrimination Object features are describing e g area shape size and homogenity The
41. vestigation object you should adjust the video settings Choose menu command Video Options and click the button Device Options To view the result of the changes immediately the video system should be in live mode continuous grabbing mode The options available depend on the imaging device connected to the system Please refer to the appendix of this manual for a description of all available adjustment options Video Options File Options Select Input Source Input File FileName C temptDekugWwideos Demo avil ae Device Options Capture Options start Capture In the Video Options dialog you may also record video sequences to a file with optional video compression using the buttons Capture Options and Start Capture If Video File is selected as an input source you will have to choose a video input file below Input File 11 meldea LAB IMAGING SYSTEM 4 Setting of system parameters To ensure optimal image processing by the medeaLAB system you should perform pixel calibration 4 1 set the gray color thresholds for object detection 4 2 and object feature filters 4 3 4 1 Pixel calibration 1 Freeze video if in live mode menu Video Snapshot toolbar i 2 Choose Options Calibration from the menu The mouse cursor will change to crosshair shape 3 Accurately draw a line between two points in the image of known distance e g cell of known diameter or a millimeter scale while
42. y are calculated continuously during image processing Using filters for these object features contaminations may be excluded from the analysis by setting the upper and lower limits appropriately For setting choose Options Form Parameter from the menu 15 IMAGING SYSTEM Object parameters Area area of an object calculated on the base of its contour in calibrated units Formfactor Ratio calculated from object Formparameter Object Nucleus Parameter Model Object Parameter circumference and area expressing the Parameter Min Acton Max Action shape or H Area 100 P s00 P roundness 1 0 means perfect circle without unit I i 7 m Length lenght of object longer main Pl width axis in calibrated units Width width of object longer main axis C Color Ava 00 00 00 FF FF FF in calibrated units L Homagenity Oj if Color Avg Mean color value of all C Convexity O gt ij object pixels C Appendis 5 Homogenity uniform distribution of color gray values inside object calculated from variance values without unit Convexity ratio of object area and area of Exclude its minimum convex hull polygon without We Hit Boundary unit Appendix presence of processes e g flagellum unit is sensitivity of detection Default Load Save Increase the sensitivity if not all of the objects having an appendix are detected Exclude Hit Boundary If this option is checked ob
43. your objects Contour Simple contour objects Contours are the base for subsequent analysis Edge Object Objects consist of corners connected by straight lines Ellipse Contours are analyzed and fitted to ellipsoidal elements This mode is used for approximation of plant leaves Skeleton Objects are skeletonized i e C Contour Segment 7 18 IMAGING SYSTEM area is decreased until only a line remains one pixel thick Based on this skeleton object length and curvature may be evaluated This option is best suited for measuring filamentous structures or fibers Contour Segment Contours are divided into convex and concave segments depending on Object Parameters above meldea LAB IMAGING SYSTEM 5 Selection of image regions All image processing functions may be limited to selected image regions Regions Of Interest ROI Depending on your field of application you can choose from six different ROI types Menu Image ROIs Symbol on toolbar Rectangle E Parallelogram S Ellipse O Free Hand o Ring Extra ROI special ROI type useful only in multiple camera configurations Delete active ROI x Note You may draw multiple ROIs in one image see picture below All ROIS are saved and automatically restored if a new image is loaded create a rectangular or elliptic ROT Press the left mouse button at the upper left position in the image window hold it move the mouse cursor to the bottom ri
Download Pdf Manuals
Related Search