Selection algorithm of Gabor Kernel for face recognition

Because the fact that Gabor feature are redundant and too high-dimensional, appropriate feature dimension reduction appears to be much more necessary. To address this problem, a novel optimal selection method of Gabor kernels´ scales and orientation is proposed. In this method, all training samples are convolved with each Gabor kernel. Within-class distance and between-class distance calculation are performed on these convolution results, respectively. At last, the optimal Gabor kernel is selected based on the ratio of the Within-class distance and the between-class distance. The Gabor Kernel corresponding to the largest ratio is the optimal one. To prove the advantages of proposed method, extensive experiments are conducted on popular face databases such as YALE, AR, FERET. The experiment results shows that the proposed method is effective and the features in the larger scales as well as the features in several orientations have more discriminative power.