APHIS User Manual
Contents
1. Photo 2 MATCHING When pre processing is finished the matching phase begins by pressing the Search or Search Batch button depending whether a single image or multiple images have been pre processed The software will calculate similarity indexes for each sample candidate pair If images are processed by batch results will be presented only when all samples have been analysed The matching phase is automatic and does not need supervision This is true for single as well as multiple images APHIS will display the results for each image sequentially VISUAL INSPECTION For each image APHIS displays in the lateral panel the first fifty options considering their likelihood values Users can browse them from highest to lowest scores to visually compare the sample and candidate After the inspecting phase the image can be registered using the Match button if a match has been accepted or as a new individual with the New button if a match has not been found User can also discard the image using the Discard Sample button Photo 3 A match occurs when the sample and the selected candidate images are from the same individual The sample image will be stored in the subfolder of the matching candidate in the database or on a new subfolder if it belongs to a new individual In APHIS each subfolder represents one individual The visual matching task will be then repeated for each sample image APHIS Automatic Photo Identification Suite Freely availab2. The second procedure also implements a database filtering system that uses qualitative parameters to avoid unnecessary comparisons Both procedures share a common user interface the characteristic of which are described in the following sections The photo identification consists in the comparison of a sample i e a picture of an individual of unknown identity against a set of candidates i e pictures of individuals of known identity present in the database The database in this case is a simple file structure that uses folders to store any images of each individual In APHIS the database will be a folder with as many sub folders as individuals identified The software assists photo identification by applying algorithms to calculate similarities among the sample images and the candidates returning a score value per comparison that will be used to order candidates for a posterior visual inspection During the final inspection the user decides whether the sample is a match with a candidate or not If the sample is a match the image will be stored under the existing folder of the candidate picture Alternatively it will be stored as a new individual creating a new folder in the database Note that APHIS prompts the folder name using successive numbers but users can write a different one The photo matching is done in three steps or phases the pre processing phase the matching phase and the visual inspection of results 1 Pre processing
3. de i Fundaci Balear d Innovaci i Tecnologia Contents Disclumerioosmcoitem s mmmensssmemteamn emtssonmesmt s tmnesnmesmtosene 2 Project partners and ACKNOWL SAS CINCHIS x5 5 esmoiiimessssttsmmmnmsntmmnsacitsmmda 3 TMOG ION a tl min REEE E a REE 5 2 HOW to Deg cele ae 6 3 Image Template Matching 11M accssicssciavessncetesensntesnanenaccnintnesvenavceoasexaniusnacinacdsunvenssnensssaberesuarineennes 7 4 Spots Pattern Mac ie CSP ccsasivesssdnancestssnsedvonmieenandncsesssupnearannnedsnstocdsesinmnesudueneavisandeaants tuxteks 12 5 Tracking the reslis oesie ae RE EA EAO EEE ER EEE 19 6 Update the database for advanced and ventured USETIS cccceesseceseceteceeeeeeseeceseceeeeeeeeeseeeaeenes 19 Te Me ATW TS AMMA Ei ii mei into 19 APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS ia i re A Fundaci Balear d Innovaci i Tecnologia 1 Introduction This manual guides you through the use of APHIS v 1 0 This software suite includes two photo identification procedures 1 a pixel based comparison of a given region or ROI Image template matching ITM 2 a comparison of user defined spots pattern Spots Pattern Matching SPM The first procedure has been originally created for APHIS whereas the second procedure is an extension of the already existing procedure implemented in software I3S see chapter 4
4. Before comparisons users are asked to prepare samples by selecting reference points on each picture This phase is different in SPM and ITM see below 2 Matching During matching the software compares the sample pre processed images with those already presented in the database and delivers a series of potential matches maximum an hundred potential matches from the more to the less plausible 3 Visual inspection In this last phase users inspect the potential matches and decide whether a match has been found or not In SPM we implemented a database filtering system which allows the user to define a filter e g male of female to reduce the set of valid candidates avoiding unnecessary comparisons against the whole database Filters have been thought to describe qualitative characters or features that distinguish unequivocally one individual from another such as the gender or the location User can define as many filters as necessary that will be combined to preselect the candidates that have to be compared to the sample We have called this functionality multichar filtering The use of such filters is not necessary but in large database can reduce considerably computing efforts and waiting times during the analysis APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS APHIS store
5. Fundaci Balear d Innovaci i Tecnologia minimum This makes APHIS especially suitable for the analyses of a large number of photos Also it includes two different routines the ITM and the SMP to better meet users needs and a multi character filtering for a more efficient search in large database In the ITM routine Chapter 3 pre processing phase is reduced to the setting of two references For expert as well as new users this per processing task would take 2 3 seconds per image However the routine has some pitfalls Deciding two reference points that can be easily recognized in any picture is not an easy task It is helpful to have a little knowledge of the anatomy of the species under study and find out a reference system unequivocally and easily recognisable in all samples A second problem is that ITM performance is highly affected by image quality and diligent usage Heterogeneous sampling conditions in luminosity focus or other distortions as dirt over the animal can affect the analysis An inconsistent setting of references has also dramatic impact on the result Therefore ITM has to be taken as a rigid procedure and cannot be considered as a definitive solution for all ecological studies Despite these drawbacks when image characteristics are homogeneous across the samples and two body references are easily recognised ITM is a very fast an efficient method for photo matching The second procedure SPM uses the existing I3S al
6. Tecnologia 5 Tracking the results APHIS stores the results in text files These files are formatted as comma separated CSV to allow its inspection on a spread sheet CSV files can be opened by commonly used programs as Microsoft Office or Open Office among others In some cases the program will open a menu asking you to rename a file Note that repeated file or individual names are not allowed in APHIS to avoid confusion during results inspection Under the project folder you will find a results logs subfolder containing four files namely Chains csv Dates csv DatesAndChains csv and Rename csv Chains csv contains a row per individual with its name in the first column and sample names that belong to it in adjacent columns ChainsAndDates csv includes also sampling dates of each sample Dates csv only the dates And Rename csv a history of renamed samples to let you track denomination changes in filenames Under the results subfolder is also created a registry for each analysis run they provide a detail of scores obtained per sample candidate pair displayed in the results lists ChainsAndDates csv includes also sampling dates of each sample Dates csv only the dates And Rename csv a history of renamed samples to let you track denomination changes in filenames 6 Update the database for advanced and ventured users Frequent users of APHIS might be tempted to manage the repository and capture folder contents This can be dangerous
7. because deleting some files can corrupt the whole project A first GOLD RULE is to make a full copy of your project folder before doing any change to its contents Keep this copy in a safe place and if something goes wrong simply substitute the corrupted folder by this backup Files can be moved across subfolders if for example a sample was incorrectly matched with an existing individual or one can delete a sample that is producing mess in your analysis etc However if these actions are made while APHIS is open during pre processing or visual inspection the software will not be aware of the changes and can react unexpectedly To avoid this one should use the Update database function in the Tools menu Once updated the program will know about the changes and behave correctly You can also edit the metadata file contents but we do not recommend to do that as far as it is easy to edit them reopening the sample inside APHIS and using the interface selectors to check or uncheck a value 7 Final remarks APHIS is a new freely available software for individual recognition by photo identification The main aspect that distinguishes APHIS from other routines is that it reduces supervised tasks to the 5 It can be downloaded for free at http www openoffice org APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS P A re A
8. been applied to detect such circumstance Analyse of image s Once a project with at least one sample in the database is open it is possible to proceed to the matching analysis The three phases of pre processing matching and visual inspection post processing are similar to the ones described for ITM However the supervised tasks are different PRE PROCESSING Users should first open a sample with the Open button or set of samples with the Open Batch button Photo 5 ME APHIS Automatic Photo Identification Suite Project Manual D proyectos IBIT FotolD FotoID2 0perativa implementacion Testing TestingFotol DER File Tools Help General Info Features Project Information Project name Manual Analyzed with 135 Image Information Name IMG 1686 JPG Date 2010 10 05 13 12 08 Finger Print Information Number of References 0 Number of points 0 2 images in Capture Folder IMG_1296a IMG_1297a Photo 5 Three references should be marked on each image These are positioned clockwise by the first three APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS di i Fundaci Balear d Innovaci i Tecnologia left click of mouse Each reference is labelled to help in this process Subsequently user should mark with the left click of the mouse spots scars or other
9. project is created the user is asked to specify the name of the project and the procedure that will be used in photo matching ITM or SPM MB New Project Project Name Analyze with Image matching 1 It can be downloaded for free at http www openoffice org APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS ia i re A Fundaci Balear d Innovaci i Tecnologia Differences between these procedures are outlined in chapters 3 and 4 The ITM procedure is faster but is more affected by image characteristics It should be preferred when images are homogeneous in terms of light and resolution and when individuals present highly defined spots or contrasted colors The SPM procedure is well adapted when photos are heterogeneous or when animal characteristics are not well distinguishable however the pre processing phase is usually longer than the one in the ITM procedure Once the matching procedure has been defined a project sub folder is created and stored in the directory ProyectosFOTOID or ProyectosI3S under ITM and SPM procedures respectively APHIS will create a project file with pis extension a database sub folder where the candidates will be stored a sub folder where edited samples will be placed since processed and a last sub folder that will contain the reports and logs generated during t
10. characteristics that are used for individual recognition Images are then automatically taken as repository The pre processing phase can be stopped at any time Edited samples will be temporally stored in the project s Captures folder ready to be analysed The General Info tab has a section that lists the samples contained by the Captures folder and a button to delete them if this is your decision It also contains an Image Information section that displays the name of the image file and the date and time of capture and a Finger Print Information section informing of the number of references and spots already marked on the active sample The pre processing phase in SPM procedure is longer than the one in the ITM depending on how many spots scars or other characteristics have to be marked MATCHING When finished with the preprocessing of the samples users can press the Search or Search Batch button the right option depends of the Open option you used one step before The software will calculate similarity indexes for each sample candidate pair and display in the lateral panel the first fifty options considering this likelihood values VISUAL INSPECTION after the matching phase APHIS delivers the results from from highest to lowest scores Here users can visually compare the sample and candidate pair Photo 6 Once taken a decision it can be registered using the Match New or Discard Sample buttons Match means that that both the sample and t
11. d function of OpenCV that slides a template image patch over an input image looking for matches This method provides three different algorithms and their normalized versions in order to calculate a matrix of likelihoods of match per comparison APHIS implements the normalized version of the correlation coefficient algorithm which is the most accurate of the three Normalization is used to minimize the effect of lightning differences among template while calculating matching scores During the ITM pre processing step the user selects two reference points for each picture They should be easily recognizable parts of the animal e g shoulders or eyes or species characteristic 2 http opencv org APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS sa A re A Fundaci Balear d Innovaci i Tecnologia For the lizard in Photo 1 the two references were the two upper corners of the larger scales on the first four scale horizontal pectoral line Photo 1 1 Depicting a reference system is necessary to ensure accurate two dimensional comparisons of sample and candidate The reliability of matching scores will depend on the reproducibility of this selection across pictures References will be used in ITM to align the individual with the horizontal axis Photo 1 2 crop a region of interest which proportions will be dependent of the pix
12. d to this area as you progress in the analysis Create the project database When creating a new project as photo identification is based on the use of a minimum repository of images to compare your sampled individuals with it is mandatory to create a database to proceed to the next steps see end of Chapter 2 for how to begin database Once an image is displayed as a single image or as the first one of a batch it appears on the main window User can control its zoom level and position either with the mouse wheel or with the size selector of the toolbar and the scroll bars that surround the image The next two left click of the mouse will position the references Photo 1 1 The first reference to set must be the left one the second the right one The sequence is not trivial as APHIS needs to know which one is on the left side of the animal and which is on the right A reference can be deleted right clicking on it After setting the references user can switch to the handy buttons from the toolbar 2 Es sze s HM OQ 7 At this stage with a new project where no repository or candidate images are present images should be simply saved in the database Press the Save button with the disk image to open one image or the Save amp Next button the disk image with a green arrow if you are in a batch process Both store the pattern and templates in the database You can also discard an image if it does not achieve the necessary quality to be
13. e aoe Ate 3 ed it Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS MEDEA Ce Fundaci Balear d Innovaci i Tecnologia APHIS User Manual Draft 9 May 2013 Automatic Photo Identification Suite Version 1 0 0 Fundaci Balear d Innovacio i Tecnologia Govern de les Illes Balears Population Ecology Group Institut Mediterrani d Estudis Avan ats CSIC UIB Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS Disclaimer This program is free software you can redistribute it and or modify it under the terms of the GNU General Public License as published by the Free Software Foundation either version 2 of the License or at your option any later version The software was developed with greatest care and tested extensively on different computer systems We found no errors in the released product versions Nevertheless there is no guarantee that the software will run error free on each system We will not take any liability for incompatibilities with hardware components and other software products and their components The software is provided as is without any guarantee on the suitability for a certain case of application No liability whatsoever will be accepted in case of data loss caused by using the software Using the software is at the sole risk of the user Fundaci BIT and IMEDEA will take no liability for damages resulting from the direct or ind
14. ecome the most active user and reviewer of the APHIS 1 0 functional beta version having analysed at present more than 3000 pictures of lizards He is currently living in front of a screen with a lizard picture The Fundacion BIT formerly Fundacion iBit has been in charge of software programming Dr Oscar Moya supervised the tasks of a team of developers ensuring that APHIS functionality fulfilled the requirements defined by the field study Pep Lluis Mansilla and Sergio Madrazo worked hard transforming ideas to the working piece of code now denominated APHIS and freely available at http imedea uib csic es bc ecopob The team made the effort of making APHIS flexible and user friendly As it is it can be used in many other species A use on the individual recognition of Majorcan Midwife toad Alytes muletensis is currently under study Moreover APHIS can be customized to meet specific needs or solve potential constraints For example additional routines can be added and preliminary statistical analyses can be automatically performed from the log files if necessary Finally APHIS would not be what it is without the work and support of Jurgen den Hartog and Renate Reijns authors of I3S who have kindly and rapidly responded to all our questions on I3S code and method APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS
15. el distance among references Photo 1 3 resize the image to a fixed resolution being the resulting image named as pattern Photo 1 4 and finally to divide the pattern in six non overlapping contiguous templates extracted automatically from the lower half of the pattern Photo 1 5 Steps 2 to 5 are automatically performed by the software and do not need supervision The only supervised task is marking the two reference points 460 px Be el 5 3 gt p z E 101 Xpx p Xpx o 436 px ei 1 2 3 4 9 Photo 1 The six resulting templates extracted from a single sample are individually compared with the candidate pattern and the scores resulting from the six comparisons are added up to produce the final sample candidate matching score APHIS delivers an ITM score list per comparison where candidates are ordered from highest to lowest by their matching likelihood the final score Individual template scores range from 1 to 1 this being the score obtained when the template is a portion of the own input image Therefore a value of 6 would be a perfect sample candidate match Using non overlapping templates the effect of local image defects has less impact on the final score enabling real matches to be well positioned in the score list ahead of random ones Creating a new or opening an existing ITM project A new project can be created from the menu bar toolbar or with the Ctrl N key short cut A windo
16. es as when creating the database Two references are positioned by the first two left click of the mouse Users should take in mind that these references should be the same on every picture The pre processing phase can be stopped at any time Edited samples will be temporally stored in the project s Captures folder for the matching processed After the matching process these temporary files will be moved to individual sub folders of the database or discarded The General Info tab has a section that lists the samples contained temporarily in the Captures folder and a button to delete them if this is user s decision It also contains an Jmage Information section that displays the name of the image file and the date and time of capture and a Finger Print Information section informing of the number of references already marked on the active sample Photo 2 APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS ME APHIS Automatic Photo Identification Suite Project Manual D proyectosiBIT FotolD FotolD2 Operativa imple DER Fie Tools Help TI seis El EH Fos fa OI 7 Project Information Project name Manual Analyzed with ITM Image Information Name IMG 1296a jpg Date 2010 04 10 12 25 14 Finger Print Information Number of References 0 1 images in Capture Folder IMG_1143
17. g the new project sub folder is under the directory Proyectos 13S The file has the pis extension a database sub folder where the candidates spots patterns and in contrast to the ITM procedure multichar filtering features will be stored a sub folder where edited samples will be placed since processed and a last sub folder that will contain the reports and logs generated during the analysis As stated before it is useful to browse this folder sub folder structure using the 4 Van Tienhoven A M J E Den Hartog R A Reijns and V M Peddemors 2007 A computer aided program for pattern matching of natural marks on the spotted raggedtooth shark Carcharias taurus Journal of Applied Ecology 44 273 280 APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS File Explorer from Windows for a better understanding on how APHIS manages the files used in the project An existing project can be opened from the menu bar tool bar or with the Ctr A key short cut SPM projects are stored with the pis extension and found under the APHIS Proyectos I3S folder Once the project is opened a lateral panel is displayed in the interface It has two tabs the General Info tab and the Features tab At this time only the project name and the procedure used for the analysis are indicated in Gene
18. gorithm which is based on pair wise comparison of the distance between points defined by the user an three references points The pre processing phase is about ten times longer than in the ITM and depends on user s experience and on the minimum number of points that need to be compared On the other hand it does not rely on photo quality or characteristics as the ITM procedure As a rule of thumb we advice to work with ITM only when luminosity definition and image quality can be standardized APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob
19. he analysis Project name and characteristics will appear in the information displayed on the right part of the window A new project is a tabula rasa where no information has been provided to the software to perform any analysis As photo identification is based on the use of a minimum repository of images to compare your sampled individuals with it is mandatory to create a database to proceed to the next steps When only one image has been taken it is possible to create a database using the Open function from the File Menu Toolbar or simply by pressing Ctr O However users commonly begin with a set of images of individuals In this case one should use the Open Batch function from the File Menu Toolbar or simply by pressing Ctr B In each case APHIS will ask to select the folder that contains the image s Note that each processed image will be separated in a sub folder representing a single individual Once the first image is displayed user can control its zoom level and position either with the mouse wheel or with the size selector of the toolbar and the scroll bars that surround the image After these preliminary steps it is time to define the references for each image Here the procedures differ and the steps to follow will be outlined in the next chapters 3 Image Template Matching ITM Introduction The ITM procedure implements the matchTemplate function of the Open Computer Vision libraries The matchTemplate is a preprogramme
20. he selected candidate are the same individual It will store the sample in the subfolder of the matching candidate in the database Therefore each subfolder represents one individual Pressing New you indicate that no candidate matches with the sample and that it should be stored as new individual Finally one can discard a comparison if you don t want to include the sample in the database This supervised task is equivalent to the one in the ITM procedure and it bound to last the same time E APHIS Automatic Photo Identification Suite File Tools View Help Test 135 8 is s Sze 100 s H H E an lt Score Candidate 1 64411 D proyectosIBIT 4 D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT D proyectosIBIT i D proyectosIBIT 16 1 D proyectosIBIT 17 1 12179 D proyectosIBIT 18 1 05999 D jproyectosIBIT 1 00014 D jproyectosiBIT 20 0 891143 D proyectosiBIT y Ae anann Milono due TOTTI lt gt CEC EI 10 i 12 13 14 15 ir SCORE Q 1 64411 Include in database as Photo 6 APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS ia i re A Fundaci Balear d Innovaci i
21. ill be used as a matching feature in the matching phase In the case of the lizard we marked the intersections between scales so to define a pattern of scales used as a fingerprint Photo 4 2 with a minimum of 30 points A reference or spot can be deleted right clicking on it A final option is to define a character for the multichar filtering Photo 4 3 Multichar filtering APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GEP GRUP D ECOLOGIA DE POBLACIONS cannot be fully controlled from the interface The specific features and their valid values have to be defined in the metadata xml text file that can be found under the project folder Users can use any text editor to edit this file Open metadata xml and add any feature between the lt data gt and lt data gt tags Each feature name and valid values have to be inserted between a lt feature gt and a lt feature gt tags There must always be an opening and a closing tag per data block The same applies for the feature name and values For a better a understanding follow the next example lt xml version 1 0 encoding utf 8 gt lt data gt lt feature gt lt name gt Sex lt name gt lt value gt Male lt value gt lt value gt Female lt value gt lt value gt Unknown lt value gt lt feature gt lt data gt The previous text shows
22. irect use of the software APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS Project partners and acknowledgements APHIS development has been funded by the Spanish Minister for Economy and Innovation project number BFU2009 09359 and the Population Ecology Group of IMEDEA The development of APHIS took more than a year of programming trials and debugging work In a way it is still a working project and users are encouraged to suggest changes for further developments and report possible bugs Dr G Tavecchia from the Population Ecology Group G E P of the IMEDEA CSIC UIB conceptualized the idea motivated by the need to find an efficient and not invasive method to individually recognized lizards Dr Moya from the Fundaci n BIT formerly Fundaci n iBit made the idea reality reviewing the existing procedures in photo recognition and outlining the first structure of APHIS He put together a team of young and passionate programmers who help to refine the software The members of the G E P group were involved in software requirements and functionality definition as well as in testing phases G Tavecchia and J M Igual conducted a sequence of sampling campaigns on the Balearic lizard Podarcis lilfordi specifically designed to test APHIS Andreu Rotger Vallespir has b
23. le at http imedea uib csic es bc ecopob GRUP D ECOLOGIA DE POBLACIONS Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia File Tools View Help TestI3S Ml APHIS Automatic Photo Identification Suite cles of ia Z LD sive 100 S Score Candidate 1 64411 D proyectosIBIT A 1 50312 D proyectosIBIT 2 1 48472 D proyectosIBIT 1 45347 D proyectosIBIT 4 1 4334 D proyectosIBIT 1 40647 D proyectosIBIT 1 32787 D proyectosIBIT 1 30567 D proyectosIBIT 1 30135 D proyectosIBIT 1 27605 D proyectosIBIT 1 25642 D proyectosIBIT 1 2519 D proyectosIBIT 1 22554 D proyectosIBIT 1 17018 D proyectosIBIT 1 15288 D proyectosIBIT 1 13872 D proyectosIBIT 1 1349 D proyectosIBIT 1 12179 DifproyectosiBIT 1 05999 DijproyectosiBIT 1 00014 D proyectosIBIT 20 0 891143 D proyectosIBIT qe Padanersa she att a anna 5 SCORE Q 1 64411 Include in database as Photo 3 4 Spots Pattern Matching SPM Introduction SPM is based on the algorithm implemented in the freely available software PS In this case during the preprocessing of a sample picture the user delimits the region of interest with three references and marks the set of spots within that will be compared during matching the defined area The coordinates of each spot in the space delimited by the references form a fingerprint like
24. onding tab of the lateral panel During edition simply check or uncheck feature values for each individual gt aE 4 d Update database 4 Search with multichar After setting the references user can switch to the handy buttons from the toolbar wf Ge Size 75 gt H HR ee aO Y At this stage with a new project where no repository or candidate images are present user should press the Save button or the Save amp Next button when in a batch process Both store the pattern and templates in the database Users can also discard an image if it does not achieve the necessary quality to be taken in consideration for the analysis Keep doing that with all the images you opened APHIS will give automatically a name to the individual you save this way however it is possible to edit this name before proceeding to the next image or finishing the process APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob GRUP D ECOLOGIA DE POBLACIONS Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia ME New Individual You are going to insert a new individual in the repository Name individual Warning while creating the database using Open Batch it is required that each image belongs to a different individual Two photographs from the same individual would be treated as belonging to different animals because matching algorithms have still not
25. possible three spot pairs as references for different affine transformations and not only those defined by the user quick search Score values are calculated for each transformation and a subsequent comparison with the candidate and the lowest is kept as final result Exhaustive searches although computer resources consuming proved to be far more accurate than simple quick searches In APHIS the pre processing and the matching phases occur separately This permits to process sequentially a group of samples and then launch the matching calculations for the whole set Once finished the automated matching the user is presented with a list of sample candidate alternatives ordered from lower to higher matching scores If multiple pictures from the same candidate are available APHIS only shows this with the lowest score Creating or opening a SPM project Creating and opening project can be done as explained for the previous procedure A new project can be created from the menu bar toolbar or with the Ctrl N key short cut A window will open asking you to give a name to the project and to choose the analytical procedure to apply In this case the 3S option should be selected ME APHIS Automatic Photo Identification Suite a61 Tools Help DE New Project Ctr N am D Open Project Ctrl 4 Haano Y Open Batch Ctrl B Exit Ctrl Q ME New Project Project Name Analyze with EE The APHIS folder containin
26. ral Info but more valuable information and functionality will be incorporated to this area as you progress in the analysis The Features tab is also empty in a newly created project but will be used to display and control your choices if you decide to apply multichar filtering to the analysis Equivalent to the ITM procedure it is mandatory to create a repository of images to use as set of candidate for photo matching Once an image is displayed as a single image or as the first one of a batch it appears on the main window User can control its zoom level and position either with the mouse wheel or with the size selector of the toolbar and the scroll bars that surround the image It is now time to define the references for each image In the ITM procedure the next two left clicks on the image would set the references In the SPM procedure three references are needed The next three left click of the mouse will set the three references sequentially and in a clockwise order Photo 4 1 Each reference is labelled to minimize mistakes General Info Features 1 El Sex Male Female Unknown ES Ciscriminant feature LI value 1 Ll Value 2 v Unknown 1 2 3 Photo 4 Note that three references have been chosen to match the requirement of the study however APHIS can be costumized to have more if necessary Once the three references are set the spots pattern is applied Here user selects each spot that w
27. s the results in text files These files are formatted as comma separated CSV to allow its inspection on a spread sheet CSV files can be opened by commonly used programs as Microsoft Office or Open Office among others The APHIS software also includes the particular concept of project which is used to define a study of a single species or population in determined conditions i e from the same geographical location if migration is implausible or studied during a segment of time or analysed with a particular algorithm ITM or SPM Finally the program offers two batch processes The first one can be activated to open a set of samples for pattern edition Open Batch and the second one to compare among each of the previously edited samples and the candidates already in the database Search Batch 2 How to begin When APHIS is launched a main window appears The main Menu is divided in three categories File Tools and Help and gives access to APHIS controls The submenus also display key shortcuts to most operations A set of short cuts to the most common operations Only lightened buttons are active in a given moment ab En IH Aa are gt On the File option users can chose between opening an existing project or create a new one ME APHIS Automatic Photo Identification Suite a67 Tools Help BS New Project Ctrl ner 21 ll Open Project Ctrl HN Gen in LY Open Batch Ctrl B Exit Ctrl Q When a new
28. statistic During the analysis the spatially corrected cloud of spots from the sample is compared with the fingerprints stored in a repository This correction is the result of an affine transformation of the sample pattern mapped onto that of the candidate one Matching scores are calculated as the sum of metric distances between spots from every pair created in a sample candidate comparison divided by the square of the total number of spot pairs Lowest scores point to likely matches being the number of spot pairs used during calculation relevant to the resulting score value although it is not yet clear to what extent Numbers in the range of 20 to 40 spots would in general ensure enough pairs for the matching however we recommend you to do some preliminary testing with a reduced set of samples to determine the number of spots that will actually require your study The matching algorithm used by SPM approach in APHIS was directly extracted from the PS Classic source code respecting its license agreement GNU Public License v2 and in accordance with its developers APHIS uses the exhaustive search version of the comparison algorithm 3 http www reijns com i3s APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS described at Van Tienhoven et al 2007 The exhaustive search uses every
29. taken into consideration for the analysis Keep doing that with all the images you opened APHIS will give automatically a name to the individual you save however it is possible to edit this name before passing to the next image Note that APHIS will treat all these unmatched images as belonging to different individuals Users should make sure that each APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia GRUP D ECOLOGIA DE POBLACIONS image saved as candidate belongs to a different individual E New Individual You are going to insert a new individual in the repository Name individualt Warning while creating the database using Open Batch it is required that each image belongs to a different individual Two photographs from the same individual would be treated as belonging to different animals because matching algorithms have still not been applied to detect such circumstance Analyse of image s Once at least one sample is in the database it is possible to proceed to search for matching images There are three steps for the analysis of sample image s the pre processing the matching and the phase of post processing or visual inspection PRE PROCESSING First open the sample with the Open button or a set of samples with the Open Batch button Each image should be pre processed by marking the referenc
30. the basic contents of a typical metadata xml file By default each project has a predefined Sex feature with male female and unknown as valid values It is mandatory to include the unknown value because phenotypic characters are not always black and white If you want for example to introduce a new feature as skin pigmentation with values bluish and greenish edit the file in the following manner lt xml version 1 0 encoding utf 8 gt lt data gt lt feature gt lt name gt Sex lt name gt lt value gt Male lt value gt lt value gt Female lt value gt lt value gt Unknown lt value gt lt feature gt lt feature gt lt name gt Skin_pigmentation lt name gt lt value gt Bluish lt value gt lt value gt Greenish lt value gt lt value gt Unknown lt value gt lt feature gt lt data gt APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob GE Cl od GRUP D ECOLOGIA DE POBLACIONS Govern de les Illes Balears Fundaci Balear d Innovaci i Tecnologia Warning Do not use Special symbols or spaces on name or value denominations If you want to separate words in those names use an underscore Neither delete the lt xml version 1 0 encoding utf 8 gt header Not following this instructions can lead to unexpected behaviours of the multichar filtering Once edited this file reopen APHIS and you will see the features and values in the corresp
31. w will open asking you to give a name to the project and to chose the analytical procedure to APHIS Automatic Photo Identification Suite Freely available at http imedea uib csic es bc ecopob Govern de les Illes Balears GRUP D ECOLOGIA DE POBLACIONS sa A re A Fundaci Balear d Innovaci i Tecnologia apply In this case the Image Matching option should be selected The ITM new project sub folder is placed under ProyectosFOTOID There you will find a file with the pis extension a database sub folder where the candidates will be stored a sub folder where edited samples will be placed since processed and a last sub folder that will contain the reports and logs generated during the analysis Once familiar with the analytical procedure it can be helpful to browse this folder sub folder structure using the File Explorer from Windows That can give you a better understanding on how APHIS manages the files used in the project thus enhancing your user experience and helping you to use the software more efficiently An existing project can be opened from the menu bar tool bar or with the Ctrl A key short cut ITM projects are stored with the pis extension and found under the APHIS ProyectosFOTOID folder Once the project is opened a lateral panel is displayed in the interface At this time only the project name and the procedure used for the analysis are indicated but more valuable information and functionality will be incorporate
Download Pdf Manuals
Related Search