Abstract: Face Detection plays a very important role in many applications, and precise localization of facial feature points is one of the important steps in face recognition. A improved method is proposed by expanding from single partial template to multi-template, and setting the local template according to human eyes and mouth. In the matching process, the corresponding template is selected to search for the matching template combined with the global template to locate the feature points. Experimental results show that the improved multi-template ASM algorithm has been greatly improved than the standard ASM algorithm, and the effect is significant and can be well applied in face recognition algorithm.

Key words: ASM(active shape model), multi-template, face detection, face recognition