Palmprint Recognition Using Directional Representation and Compresses Sensing

In this study, based on directional representation for palmprint images and compressed sensing, we propose a novel approach for palmprint recognition. Firstly, the directional representation for appearance based approaches is obtained by the anisotropy filter to efficiently capture the main palmprint image characters. Compared with the traditional Gabor representations, the new representations is robust to drastic illumination changes and preserves important discriminative information for classification. Then, in order to improve the robustness of palmprint identification, the compressed sensing is used to distinguish different palms from different hands. As a result, the palmprint recognition performance of representative appearance based approaches can be improved. Experimental results on the PolyU palprint database show that the proposed algorithm has better performance and with good robustness.