Iris Recognition Using Supervised Learning Based on Matching Features

Biometrics is a discipline that studies methods for verification and identification of individuals based on physical or behavioral characteristics of a person. In this paper, an iris recognition system is proposed using supervised learning based on statistical features of matching points obtained from speeded up robust features, which are invariant to transformations. Our system extracts statistics from correspondence patterns between the pair of iris images to be compared to generate an efficient feature vector. A set of these feature vectors, obtained from several iris samples, feeds a learning algorithm to automatically classify whether the pair of images corresponds to the same iris. To evaluate the recognition rate of our system, we performed experiments on the CASIA iris image database, obtaining a recognition rate of 99.94%, with a False Acceptance Rate (FAR) of 0.00%, and a False Rejection Rate (FRR) of 0.12%, which shows efficient classification rates; moreover, this model achieves the fastest computational (1.18 s) time compared to other iris recognition methods.