Improved iris segmentation based on local texture statistics

High performance human identification using iris biometrics requires the development of automated algorithms for robust segmentation of the iris region given an ocular image. Many studies have shown that iris segmentation is one of the most crucial element of iris recognition systems. While many iris segmentation techniques have been proposed, most of these methods try to leverage gradient information in the ocular images to segment the iris, rendering them unsuitable for scenarios with very poor quality images. In this paper, we present an iris segmentation algorithm, which unlike the traditional edge-based approaches, is based on the local statistics of the texture region in the iris and as such is more suited for segmenting poor quality iris images. Our segmentation algorithm builds upon and adapts the seminal work on Active Contours without Edges [6] for iris segmentation. We demonstrate the performance of our algorithm on the ICE [2] and FOCS [1] databases.