A generic computer platform for efficient iris recognition

This paper presents a new open source MATLAB based generic platform for the development, testing and evaluation of iris recognition algorithms. This platform uses industry standard image datasets and test methodologies to evaluate the performance of iris recognition algorithms and output consistent, comparable and quantifiable results. The results generated by the platform include industry standard FAR vs FRR, ROC and EER as well as true match, false match and separability. This improves the ability to determine the efficacy of a given algorithm or variation. This presentation shows existing open source algorithms being adapted to work within the platform and then evaluated using the platform with openly available iris image datasets. The simplicity and speed of adaptation and the comparability of the results produced is clearly demonstrated. Furthermore, the benefits of using the platform for algorithm development and improvement are highlighted by demonstrating the platforms ability to interchange algorithms, or parts of algorithms with the click of a mouse.