Skin-active shape model for face alignment

Active shape model is considered as one of the most useful methods for face alignment. In this paper, we propose using a skin color likelihood transformation method, instead of using a gray level, to model the local profile around each landmark point of active shape model. The goal of using this method is to model the local profile more accurate and to decrease the influence from different races and lighting conditions. To solve the detail losing phenomena problem when using the skin color likelihood transformation method, combination of skin color likelihood and gray-level in multi-resolution is suggested.