Study On Iris Localization Alogrithm

With the rapid development of modern information technology, the issue on information security is becoming more and more important. So biometrics technology has been applied in many fields. An emerging automatic biometrics system based on iris recognition will be applied widely in areas on personality identification in the future.Iris localization is an important stage in iris recognition system. Its accuracy and speed can influence the efficiency of the whole system. So when improving the speed and accuracy of iris localization, we must try our best not to lose iris texture information. In this thesis, the method based on the strategy "from coarse to fine" is introduced in detail according to the geometrical and gray features of irises.Firstly the development of iris recognition and actuality of iris localization is presented and the advantage and shortage of existing iris localization algorithm are analyzed. Secondly in the stage of coarse localization, the inner iris boundary is extracted and localized by applying binary segmentation; the outer boundary is localized by gray gradient and transcendental knowledge. Afterwards in the stage of fine localization the inner boundary is localized based on error square summation and geometrical features. To improve the precision and speed of iris localization the outer boundary is localized by applying modified Daugman algorithm. Finally the localized iris image is normalized.At the end of this thesis some existing problems are pointed out, then the good future of iris recognition technology is predicted.