An Iris Recognition System Based on Analysis of Textural Edgeness Descriptors

A novel iris recognition system is presented in this paper. Based on the studies of different textural edgeness descriptors, a well-discriminate texture feature is employed and investigated for feature computation from iris pattern. Here, the proposed system consists of three components: pre-processing, feature extraction and classification. During pre-processing, an iris portion is segmented out from an input eye image which is then normalized to extract features for segmented iris portion. Classification is the task either to verify or to identify a subject whose iris image is under consideration. Five benchmark iris databases, namely MMU1, UPOL, CASIA-v3-Interval, IITD and UBIRIS.v1, are used to test the performance of the proposed system. An extensive experimental result is demonstrated for each database and the results are compared with the existing methods. The results show the robustness and effectiveness of the proposed approach.