A robust phase information extraction using 2-D quadrature filtering (monogenic and 2D-log Gabor) and modified HD for matching

The human iris recognition system is an attractive technology for identity authentication. This technology benefits from random variations in the features of the iris. Usually, an iris recognition system has 4 modules: segmentation, normalization, feature extraction and iris templates matching. This work is mainly focused on iris texture analysis and templates matching which are 2 essential processes in the iris recognition system. The proposed approach extracts robust phase information using filtering (both monogenic and 2D log Gabor). Then, two types of distance measures such as modified HD and Jaccard distances are chosen as metrics for recognition. We comparatively evaluate the performance of the proposed method and the fractal analysis using CASIA-V3.0 iris image databases. The obtained results with monogenic and 2D-Log Gabor filters were highly promising and led to significantly improved performance in speed and accuracy. With dissimilarity modified Hamming distance; we improved the accuracy of the iris recognition system, with a FAR equal to 3% and a speed at least 8 times.