An Automatic Geometric Features Extracting Approach for Facial Expression Recognition Based on Corner Detection

This paper presents a novel automatic facial key point extraction method based on corner detection approach. A pyramid-based FAST corner detector has been developed to seek out all the corners of facial organs with a few useless corners. It was tested and compared with four feature points detectors including FAST, SUSAN, Harris and SIFT to evaluate the effect. Then, for filtering useless corner points, facial mask that generated by edge detection and morphology algorithm was performed as primary screening. And 10 organ corner key points as eyebrows, eyes and mouth were located by the position. After that, another 10 key points in the middle of organs were located according to them. Finally, facial model was constructed by Delaunay triangulation and geometric features were computed as vectors. The detecting results show that the method can locate key points accurately on both grey and color images. And it was proved to be an effective method by 200 samples from JAFFE and MMI database.