A three-dimensional muscle-based facial expression synthesizer for model-based image coding

摘要

A three-dimensional muscle-based facial expression synthesizer is proposed. The proposed synthesizer will compute the contraction of 19 muscles and rotation of the jaw from 22 feature points estimated by the analyzer, then apply the muscle contraction model to modify the 3D head model. The head motion, including translation and orientation, is also derived from a set of three feature points. Currently the analyzer is implemented on a PC with a camcorder with near real-time performance, and the synthesizer receives the feature points from the analyzer and synthesizes facial expressions on an SGI Indigo in real time.