Efficient Face Expression Generation for Face Recognition

In real-world applications, facial expression significantly affects the performance of a face recognition system. Variation of facial expressions or poses is still challenging problem even for best face recognition methods. Conventional face recognition methods like Eigenfaces do not directly consider influence different expressions. In this paper we extend idea of rule based generation of face expression and propose new approach how to create generation rules based on evolutionary algorithms. Although new samples are synthetic, they significantly improve face recognition rates. Comparison with other approaches is also presented.