Generating facial expressions based on estimation of muscular contraction parameters from facial feature points

Abstract

Muscle movement inside the face generates the real facial expressions of a human. This paper proposes a method that generates facial expressions based on estimating muscle movement, i.e.. muscular contraction parameters using detected facial feature points. First, the facial feature points of a facial image are detected using image processing methods. Then, the displacements of facial feature points, which correspond to the muscle-based facial model that Waters proposed, are calculated. Finally, we estimate the muscular contraction parameters of a facial model to obtain the vertex displacements. Experimental results reveal that our approach can generate a facial expression of the facial model, which corresponds to the facial expression of the actual facial image