Authentication System using Multispectral Palmprint

Abstract

Abstract Among the numerous biometric systems presented in the literature, multispectral palmprint authentication systems have received a great deal of attention in recent years. This paper presents applications of the Holistic methods such as Gabor + Fisherpalm space, Gabor+Eigenpalm space and Gabor Kernel Eigenpalm space on the multispectral images (Green, near-infrared, Red and Blue). The Gabor filter bank is applied for extract the augmented magnitude feature vector. These vectors are then subjected to subspace projection (i.e., Fisherpalm space). In this paper, a comparison of the Gabor+ Eigenpalm and the Gabor+ Fisherpalm methods respect to multispectral palmrint images is examined. Finally, nearest neighbor classifier and cosine Mahalanobis distance are used for respectively matching and decision stages. Experimental results show that our proposed system achieves better reults than results than other state-of-the-art system.