Iris Recognition Technology

With the development of information technology and increasing demand of security, iris based personality authorization and recognition as a high reliability and non-offensive bio-information recognition technology have been more and more important.Iris recognition system is made up by iris localization component, iris image preprocessing, iris feature extraction and classification. In this paper based on the development of current iris research results, we researched in iris segmentation, image preprocessing, feature extraction, iris code classification. In this paper, also some new algorithm improvements have been proposed and give some experiment results.In iris segmentation, we proposed a fast iris localization algorithm: before iris localization, the algorithm enhanced the iris edge contrast and suppress the iris texture. In the localization process, we use the improved Hough transformation and simple threshold decision. Improved iris localization algorithm is faster than Daugman proposed algorithm 60 times. Also based on the iris localization result, we proposed an iris quality evaluation standard.In iris texture process, we use Log-Gabor filter instead of Gabor filter to extract the texture feature and encoding the phase response with Gray code. Finally, under large experiment results, we also discussed the filter parameters which affect the decidability, and choose the best parameters as our iris recognition parameters.In iris code classification process, we use FAR, FRR and nearest neighbor decide rules as the code classification rules. In large iris database recognition, we also proposed the iris indexed-code to improve the recognition speed. Our iris recognition system finally on CASIA database achieved recognition rate 99.931 %.