Iris recognition using templates fusion with weighted majority voting

Abstract

Iris recognition is one of the most powerful techniques for biometric identification. The requirement for current systems is to acquire multiple iris codes from the same eye and evaluate which bits are the most consistent bits in the iris code. When the acquired images are noisy, the inconsistent bits in the iris code should be masked to improve performance. This paper thoroughly investigates the use of multiple training samples for enrolment. Based on this, an enhanced iris recognition approach is proposed employing the fusion of a set of iris images of a given eye using the most consistent feature data. The algorithm reduces the database size and accelerates the matching process. The Chinese Academy of Sciences – Institute of Automation (CASIA) iris database is used to simulate the studies. The approach used in the proposed work outperforms existing approaches with the fact that in the proposed iris recognition system, verification is done with a single template (final fused template) instead of multiple base templates as done by the existing works.