Ensemble-Based Kernel Fisher Analysis for Face Recognition

This paper proposes an Ensemble-based kernel fisher analysis method for face recognition, which can effectively increase the performance of the histogram of gabor phase pattern (HGPP) method. The novelty of the paper lies in that it explains in theory why histogram can be combined with kernel fisher method, which the extended Chi-square similarity rules are positive definite. We then proposed the ensemble-based kernel fisher method to enhance the performance of HGPP, experiments on the large-scale FERET and CAS-PEAL database show that the proposed method gets much better recognition rates than the HGPP.