Template-based portrait caricature generation with facial components analysis

This paper presents a template-based facial caricature generation approach. Since the major difficulty in generating facial caricature automatically is the uniqueness of individuals and the polymorphism of features, a modified active shape model (ASM) is first designed to locate key feature points accurately by using 2D local grey-level structures. Then, extracted facial components are classified into low-accuracy ones and high-accuracy ones according to their fitting accuracy and are processed respectively. Hierarchical clustering combined with the template-based method is employed for low-accuracy components to bridge the gap between the accuracy provided by feature localization and that required by caricature generation. Experimental results illustrate the effectiveness of the presented approach.