Expression recognition based on bidirectional 2DPCA and PSO

For the sake of increasing the expression efficiency, this paper proposed an approach based on the 2DPCA and PSO(Particle Swarm Optimization). Firstly, use the amplitude of the expression images after Gabor wavelets transform as original features; then use the bidirectional 2DPCA to extract the main features; at last, use a novel evolutionary algorithm PSO to search coefficients of the features to optimize the features. The experiments based on the JAFFE database perform excellent recognition results, and in comparison, the approach presented by this paper is superior to many other methods.