Image-based face drawing using active shape models and parametric morphing

This paper describes algorithms for automatic facial feature extraction by using Active Shape Models (ASM). Based on these extracted facial features, face drawing is created, which embodies the individual features that the face looks more likely to be so. Model point initialization is implemented after face detection and eye location using horizontal textures and intensity valley correlation. Then facial features are located by a search progress of several iterations to the final convergence, which gives a good match to the target feature points. For the face drawing, feature contours and skin color are processed respectively. Feature contours represented by ASM are sketched with multi-level values to mimic the brush drawing style. Skin color is rendered by smoothing process in order to get the watercolor looking effect. Finally, parametric morphing is carried out to generate several expressions that embody a few basic emotions such as happiness, sadness, anger, and surprise.