Caricature is a method to show makeable facial features by exaggeration and deformation. Though sample based learning method can capture caricaturists' style and get well caricature effect,obtaining cartoon sample database of various styles is very difficult. So this paper proposes a new caricature generating algorithm. This algorithm classifies face by the shape of face first,dial is il needs oniy a few lace sanipics,ami extracts major feaiuics or facial shape by ASM, then obtains SVMfacial shape classifier by training these features to identify facial shape. Secondly,obtain partial feature differences of object face and standard face. Finally,exaggerative deformations on facial type and partial features are carried out with a circle areabased deformation algorithm,and then obtain final cartoon face by facial illustration processing on exaggerated face. The result of this algorithm is evaluated by questionnaire investigating,and the result is satisfied.
展开▼