Investigating Palm Vein Pattern Recognition Methods

Abstract

Biometric systems for contactless identification of a person by palm blood vessels have become widespread recently. The relevance is due to a combination of such factors as hygiene, higher degree of protection against counterfeit, and delicacy compared to other methods. Vein pattern recognition algorithm can be divided into two steps. The first step, pre-processing, is to detect the Region-of-Interest (ROI). The second step involves extracting singular points of the vein bed (feature extraction) and matching the vein pattern with the database templates (feature matching) which can be performed using various methods. The paper presents the study of several parametric methods for recognizing and matching the input image with the database templates - correlation calculation, calculation of singular points (descriptors) and calculation of perceptual hashes. The comparative analysis of the results was carried out according to the following criteria - evaluating the recognition quality, evaluating the processing speed, estimating the size of the database. We have revealed a significant advantage of the hash method in comparison with other parametric methods under study within the paper. The results will be of use for developers of authentication systems based on image analysis.