Multimodal personal authentication system fusing palmprint and knuckleprint

Abstract

This paper proposes a multimodal personal authentication system which makes use of palm and knuckleprint traits. Biometric images are enhanced and robust corner features are tracked using Lukas and Kanade tracking. Matching score between feature vectors of two images is obtained through a similarity measure which makes use of geometrical and statistical characteristics. The proposed system is tested on chimeric multimodal databases created by fusing two publicly available palmprint databases CASIA and PolyU along with PolyU knuckleprint database. Experimental results reveal correct recognition rate of 100% with EER less than 0.1%.© Springer-Verlag Berlin Heidelberg 2013.