Comparative Analysis of Palmprint Matching Techniques for Person Identification

Abstract

Biometrics has been an arising field of research in the recent years and was concerned by using physical traits, such as those based on iris or retinal scanning, face recognition, fingerprints, or voices of individuals to be identified. Applying low resolution devices palmprint is easily captured also it is distinctive, compared to other methods such as fingerprint or iris palm print has preferred as well as it includes additional features such as principal lines. In past decades, various palmprint identification methods have been proposed such as coding based methods and principle curve methods. Yet, some special kinds of biometric traits have a similarity and these methods cannot exploit the similarity of different kinds of traits. This work designs a framework at the matching score level by combining the left with right palmprint. In this framework, using the left palmprint matching, right palmprint matching and crossing matching between the left query and right training palmprint three types of matching scores were calculated and these were fused to make the final decision. This framework not only combines the left and right palmprint images for identification, but also correctly exploits the similarity between the left and right palmprints of the same subject. The fusion is done by using three methods such as line based method, Scale Invariant Feature Transform (SIFT) method and Local Binary Pattern (LBP). The experimental result shows that the LBP method provides better performance in terms of recognition, false positive and negative rate compared to other two methods. Keywords: Biometrics; Multimodal biometrics; Palmprint recognition, Line based method; SIFT method; LBP method