No abstract

As a recently proposed technique, sparse representation (SR) has been widely used for pattern recognition. Sparse representation emphasizes the coefficient sparsity and ignores the importance of the collaboration between classes. In this paper, collaborative representation is introduced for palmprint recognition, and the inter-class collaboration is employed to estimate the representation coefficient with regularized least square method. The algorithm based on collaborative representation was evaluated on the Hong Kong PolyU (v2) palmprint database and ideal recognition performance was achieved.