Steerable Riesz wavelet based approach for iris recognition

In this paper, we explore the applicability of first and second order monogenic Steerable Riesz wavelet components for iris recognition. These wavelets provide powerful mechanism to extract the invariant as well as covariant local variations of iris patterns. Unlike other existing methods where sole iris (either left or right) is used for recognition, in our work, we extract the features from both left and right irises, encode them separately and perform bit level fusion. Extensive experimentation has been conducted on the benchmark databases namely, IITD, MMU v-2 and CA-SIA v-4 distance to exhibit the performance of the proposed method. Comparative analysis is also performed with the state of the art methods to justify the suitability of the proposed method for iris recognition.