Spectral Graph Wavelet Theory Based Person Authentication System Using Multispectral Palmprint Images

摘要

Person authentication plays an important role in security based day to day life activities such as access control, identity recognition, data management and forensics. In this paper, palmprint based person recognition system is proposed by exploiting Spectral Graph Wavelet Theory (SGWT). Palmprint is a kind of human physiological trait and it has significant potential and discriminative features for person recognition. To extract features, palmprint images undergo SGWT decomposition. Then all the sub bands are fused and energy is computed as feature vectors for classifier training using nearest neighbor. The performance of proposed system is investigated on PolyU multispectral palmprint database images. Experimental results show that the proposed system achieves over 98% on blue spectral palmprint images.