Palmprint Recognition Using Kernel Spectral Regression Discriminant Analysis and HOG Representation
Wei Jia, Jie Gui, Rong-Xiang Hu, Yingke Lei · 2010
In this paper, we propose a new appearance based approach for palmprint recognition, which combines Kernel Spectral Regression Discriminant Analysis (KSRDA) method and HOG representation. KSRDA is the kernel version of SRDA which has lower computation complexity than Linear Discriminant Analysis (LDA). Meanwhile, HOG representation isn't sensitive to changes of illumination, and has the robustness against deformations because slight translations and rotations make small histogram value changes. As a result, the proposed approach can achieve promising recognition rate. The results of experiments conducted on Hong Kong Polytechnic University Palmprint Database II and the blue band of Hong Kong Polytechnic University Multispectral Palmprint Database demonstrate the effectiveness of proposed approach.