Palm Dorsa Vein Pattern based biometric verification system using Anisotropic Generalized Procrustes Analysis on weighted training dictionary

Abstract

Abstract This paper proposes a novel Palm Dorsa Vein Pattern (PDVP) based Biometric system for human identity recognition. The strategy in this research work specifically aims at the robust identification of real human subjects using far infrared (FIR) thermal images of the PDVP as a physiological biometric feature. One of the most significant aspects of any biometric system is the feasibility of data acquisition and, therefore, in our research we have focused on a non-stringent setup for data acquisition. Consequently, one of the critical issues related to the infrared images of the PDVP acquired, under real environmental conditions, without any proper-and-fixed setup may result in the presence of scaling, angular, and/or translational displacement. Under such circumstances, the recognition rate may get affected. Moreover, most of the existing strategies do not address such critical issues of data acquisition. Therefore, to address such a fundamental issue that poses a stiff challenge to the recognition procedure we, in this paper, propose an algorithm for recognition which is based on Anisotropic Generalized Procrustes Analysis (AGPA) of the similarity between the test PDVP image and the weighted training samples. The weight matrix for the associated images is formulated using the Tikhonov Regularization methodology and then a Hadamard product of the weight matrix and the training dictionary is performed prior to the similarity analysis stage. Moreover, owing to the fact that a Biometric system is not just about identification/recognition but also about authentication, we have formulated an authentication strategy utilizing the same AGPA based similarity analysis. Theproposed method is tested on a well-structured database, of thermal images of the PDVP, JU-FIR-V2: FIR Vein Database , developed in the Electrical Instrumentation and Measurement Laboratory, Electrical Engineering Department, Jadavpur University, Kolkata, India . Subsequently, through extensive experimentation it has been proven that the proposed strategy attains substantially high and stable recognition rates and, additionally, performs equally well with respect to authentication of a test sample.