A complete end-to-end system for iris recognition to mitigate replay and template attack

Abstract

Widespread use of iris biometric-based authentication makes it vulnerable to several attacks like template attack, replay attack, print attack. Several approaches have been proposed to mitigate each attack individually but nothing could be found, in the literature, that handles them collectively. A complete end-to-end system is required that shall be capable to handle these attacks together rather than just focusing on a particular type of attack. In this paper, we propose a system, which is capable of handling replay attack and template-based attack and paves a path to the evolution of a complete secured system. A non-deterministic approach for iris recognition, based on robust regions, proposed earlier (Gupta and Sehgal in Pattern Anal Appl 1–13 (2018), [1]) has been used to mitigate template-based attack along with replay attack. Biometric-based key generation is one of the techniques to evade the template-based attack. It requires a key generation to authenticate the user. The robust regions are further shown here to be effective in iris key generation as well. This eludes the necessity of saving iris template and the use of biometric keys for user authentication. The entropy of our system is calculated as 57 bits which shows the effectiveness of the proposed approach.