Face Recognition Based on Modified Modular Principal Component Analysis

Abstract

The technology of face recognition has been widely applied to many fields such as identity authentication. A New Improvement for Face Recognition Using MMPCA is presented in this paper. The proposed algorithm when compared with conventional modular PCA algorithm is different in the computation of image mean value and the recognition process. Comparison of the two algorithms in different face databases proves that the proposed algorithm is more effective and robust than conventional modular PCA algorithm under the large variations in lighting direction and facial expression. The authors also point out that 2DPCA is a special case of improved algorithm, no matter in the process of dimension reduction or recognition.