A New Two-Directional Two-Dimensional Feature Extraction Based on Manifold Learning

A new feature extraction method based on manifold learning is proposed for face recognition in the paper; its criterion function is characterized by maximizing the difference between the nonlocal scatter and the local scatter. The novel method is called two-directional two-dimensional marginal discriminant projection ((2D)²MDP), which simultaneously works image matrix in the row direction and in the column direction for feature extraction. The experimental results on ORL face databases indicate that the proposed method has higher recognition rate and more stable.