A Face Recognition System Based on Kernel Maximum Between-class Margin Criterion(KMMC)

This paper proposed KMMC(Kernel maximum between-class margin criterion) as the basic extraction method of face recognition to avoid small sample size problem in face recognition. Moreover, based on the maximum of the difference between between-class scatter and within-class scatter in feature space, we obtained a set of optimal discriminant vectors as the projection axis to proceed projection transformation, so that make the between-class scatter of Kernel feature space sample maximum and the within-class scatter minimum. Theoretically solved the unanswered problem caused by singularity of within-class scatter, and further demonstrates its efficiency of feature extraction. Through the test on ORL face database, the results confirmed the validity of the feature extraction method. At last, by using Matlab, we designed a face recognition system based on KMMC.