An Improved Algorithm for Facial Feature Location by Multi-template ASM

To improve the active shape model(ASM) location accuracy in smooth regions in facial images,this paper proposes an improved multi-template ASM based algorithm which performs a coarse location by a global template at first,then followed by a local template for refinement.At the second stage,a narrow strip-map is first constructed on the feature points of each template,and then a Closed-form algorithm is employed to perform texture segmentation on the narrow strip-map;finally local templates are matched to the image to obtain feature-point information.Experimental results show that our algorithm significantly improves the location accuracy of the traditional ASM in smooth regions,and enhances the detection accuracy of all the feature points.