Face Feature Points Extraction Based on Refined ASM
基于改进ASM的人脸特征点提取

Recently ASM/AAM based face alignment is becoming a more and more interesting research topic in computer vision research community. A novel ASM based face feature points extraction algorithm is proposed in this paper, which uses gradient local features to improve alignment accuracy and lower the size of training samples and uses texture constrained shape to improve the feature extraction accuracy. The proposed algorithm has been evaluated by a number of statistical experiments. The potential applications vary from face tracking, recognition to face expression analysis.