Recognition Based on Metric-optimized Neighborhood Preserving Embedding

Abstract

Face recognition is a biometric technology with great developable potential. It has a great deal of potential applications in public security and information security. To overcome the problem in the high-dimensional face data processing, the k-nearest neighbors is chose by Linear Discriminate Analysis (LDA). A Metric-optimized is proposed for Neighborhood Preserving Embedding (MONPE).MONPE algorithm, with the dimensions of data reduced by LDA, will be reasonable in NPE algorithm. On the other hand, LDA maximizes the between-class scatter and minimizes the within-class scatter, so the neighbors of a sample will have higher possibility to be picked from the same class .With the ORL face database and the Yale database, the recognition rate and run time is compared among NPE, MONPE and CLMONPE. The simulation results show that CLMONPE has obvious advantage in application

DOI : http://dx.doi.org/10.11591/telkomnika.v12i4.4825