Palmprint Recognition Based on RB K-means and Hierarchical SVM

In this paper, we present a novel recognition algorithm which is based on RB K-means and Hierarchical SVM (Support Vector Machines). Firstly, we pre-classify imposter samples by RB K-means. Secondly, the Linear SVM is applied to the primary classification. Finally, we use the Non-linear SVM for the classification. Our experiment demonstrates that our algorithm with RBF kernel is feasible and effective in terms of FRR and FAR (FRR:0.242%, FAR:2.668%) on our palmprint database.