Iris Image Matching

C++ based iris image verification code using OpenCV

This iris verification algorithm is originally based on A Biometric System for Iris OSIRIS version 4.1 developed by Telecom Sud Paris. The original version is based on C-style OpenCV, whereas this version is based on C++. You may access the original version here:

http://svnext.it-sudparis.eu/svnview2-eph/ref_syst/Iris_Osiris_v4.1/

Before you use:

1. Make sure you have installed OpenCV on your computer, since the code relies on it.

2. Copy those header files under folders named opencv and opencv2 from your installed OpenCV location to x64/Release/HeaderFiles/ of the solution, or change the opencv header file source directly from the Solution Properties -> C/C++ -> General -> Additional Include Directories to point towards your original opencv and opencv2 paths, such as C:\OpenCV-2.4.10\build\include\opencv and C:\OpenCV-2.4.10\build\include\opencv2.

3. Create a folder as x64/Release/Libs/ and copy those OpenCV lib files under it, or change the related solution property, by changing Solution Property -> Linker -> General -> Additional Library Directories to your original OpenCV lib path, such as C:\OpenCV-2.4.10\build\x64\vc12\lib.

4. For the whole description of the iris matching algorithm and processes, take reference for the following documentation: http://svnext.it-sudparis.eu/svnview2-eph/ref_syst/Iris_Osiris_v4.1/doc/

5. The main code reads two iris images, and performs matching. The final matching result is a score between 0.0 and 1.0, where 0.0 means the two iris images are totally different, while 1.0 means they are exactly the same.

6. The parameters for Gabor filters used are saved in those files under PirisParam folder.

7. You can change relevant contents of file configuration.ini for other purposes, such as writing out the matching score result into a file, etc. Read the above mentioned documentation for further details.

8. This distribution is subject to GNU GENERAL PUBLIC LICENSE.

9. Assume that these are your two input images:

This is the final output:

Best wishes,

Ekberjan