Multispectral palmprint matching based on joint sparse representation

Abstract

A novel method for multispectral palmprint matching based on the joint sparse representation is proposed. We use joint sparse representation to model the identity assurance system that involves identification as well as verification. The method represents the given palmprint as a linear combination of the multispectral palmprints. The information from different spectrum are fused by means of feature level fusion. The nearest neighbour classification based on class wise reconstruction error is used for classification. Experiments are conducted on PolyU multispectral palmprint database. The results show that the proposed method works better in comparison with the existing techniques.