Exploiting stable and discriminative iris weight map for iris recognition under less constrained environment

摘要

In this paper, we address the problem of iris recognition under less constrained environment. We propose a novel iris weight map for iris matching stage to improve the robustness of iris recognition to the noise and degradations in less constrained environment. The proposed iris weight map is class specific considering both the bit stability and bit discriminability of iris codes. It is the combination of a stability map and a discriminability map. The stability map focuses on intra-class bit stability, aiming to improve the intra-class matching. It assigns more weight to the bits that are highly consistent with their noiseless estimations which are sought via low rank approximation. The discriminability map models the inter-class bit discriminability. It emphasizes more discriminative bits in iris codes to improve the inter-class separation via a 1-to-N strategy. The experimental results demonstrate that the proposed iris weight map achieves improved identification and verification performance compared to state-of-the-art algorithms on publicly available datasets.