An Improved Iris Localization Method for Authentication System

At present, the research of biometric recognition is gave more and more attentions. The iris recognition is a kind of biometrics technologies. Based on Daugman´ and Wilds´ iris localization algorithms, the paper is to present an improved iris localization method named two-phase approximation localization for authentication system. Firstly, we evaluate the image quality and abandon blurred pictures. And we estimate the center and radius of iris boundaries. After that, the method calculates the center and radius for the second time using vote method. The average time of localization reaches 2.0 seconds and 99.70% at accuracy. The method is much faster than Daugman´s and Wilds´ classical methods.