A Practical Iris Recognition Algorithm

This paper proposes an iris recognition algorithm based on 2D zero-crossing detection and similarity classifier. Whole system is consisting of eye image capture, iris boundary localization, iris region normalization, feature extraction, pattern match, and yes/no decision. In iris feature extraction stage, iris normal region is filtered by using low frequency filter firstly, then the texture features are extracted by using 2D zero-crossing detection operator, at last iris features are encoded into binary feature template. Iris pattern match can be computed by computing similarity degree of binary templates. Goodness-of-match measurement can be used to distinguish which class the sample belongs to or identify authentic or imposters. Simulation results show that this algorithm is effective.