Combination of Local Multiple Patterns and Exponential Discriminant Analysis for Facial Recognition

Global features-based methods and local features –based methods have been very successful in face recognition system, yet they can be combined together and jointly optimized so as to minimize the error of a nearest-neighbor classifier. We consider both descriptor for face images with Local Multiple Pattern, and discriminant learning techniques with Exponential Discriminant Analysis. A combination framework based on Local Multiple Pattern and Exponential Discriminant Analysis has been proposed in this paper. Firstly, our approach encodes the multi-scale face feature by Local Multiple Pattern, and then they have been extended to strengthen the discriminative ability by Exponential Discriminant Analysis; Secondly, we suggest to use the above feature on different layers independently so that a multiple classifier system can be attained. Using these techniques, we obtain the state-of-the-art performance on two public available databases. Copyright © 2013 IFSA.