Face Recognition Method Based on SURF Feature

Feature extraction is crucial for face recognition. A new method of face feature extraction based on Speeded-Up Robust Feature (SURF) is proposed in this paper. SURF is invariant of shifting, rotation and scaling, and partially invariant of illumination and affine transformation. We use Fisher Linear Discriminant (FLD) method to extract the quadratic features on the basis of SURF feature, then measure the similarity of faces by calculating the Euclidean distance of the quadratic features. The experiments on ORL and UMIST face databases have been carried out. The experimental results indicate that the proposed method can reach a high recognition accuracy effectively.