In this paper we propose a novel approach for generating expressive caricatures from an input image. The novelty of this work comes from combining an Active Appearance Model facial feature extraction system with a quadratic deformation model representation of facial expressions. The extracted features are deformed using the quadratic deformation parameters, resulting in an expressive caricature. The facial feature extraction requires an offline training process which uses natural expression annotated images from 30 model subjects, selected randomly from the Cohn-Kanade Database. The results show that from an input facial image, expressive caricatures are generated for the main six face expressions (smile, sad, fear, surprise, disgust, and anger). The proposed approach yields to promising expressive caricatures, and could lead to future research directions in the field of non-photorealistic rendering. In addition, the proposed approach can be employed in entertaining standalone applications or caricature animations.
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=1690437.