A hybrid face recognition algorithm based on WT, NMFs and SVM
Pei Jiang, Yongjie Li · 2008
In this paper, we proposed a new scheme for face recognition, which hybridizes wavelet transform (WT), non-negative matrix factorization with sparseness constraints (NMFs) and support vector machine (SVM) with relative difference space (RDS) method. Firstly, low frequency subband images are extracted from original face image with 2D wavelet transform. Secondly, the images with low frequency information are factorized with NMFs, which could find part-based representations of images. Then, the multi-class problem is to be converted to the binary issue by RDS method and the extracted features are classified through SVM. The experiments on ORL face dataset shows the more efficient results with the proposed algorithm.