Multiple Feature Point Discriminant Analysis and Its Application to Feature Extraction

In this paper, a novel linear subspace learning approach, named Multiple Feature Point Discriminant Analysis (MFPDA), is proposed. MFPDA is in order to maximize the multiple feature point between class scatter and minimize the multiple feature point within-class scatter. Some experiments are performed on FKP database, AR face database, and ORL face database to evaluate the effectiveness of the proposed MFPDA. Compared with some popular subspace learning methods, such as PCA, LDA, LLP, UNDFLA, JSPCA, the proposed MFPDA has highest average recognition accuracy. The experimental results confirm the effectiveness of the proposed algorithm.