人脸肌肉模型的单视频驱动  

Drive Muscle Face Model Using Single Video

作　　者：段鸿[1] 王以孝[1] 程义民[1] 蔡尚书[1] 

机构地区：[1]中国科学技术大学电子科学与技术系,合肥230026

出　　处：《系统仿真学报》2003年第9期1268-1271,共4页Journal of System Simulation

摘　　要：描述了一种肌肉人脸模型单视频驱动的方法。与传统的通过多视频来驱动人脸模型的方法不同,该方法通过跟踪特定的人脸控制点的位移,来求取人脸肌肉的运动参数,进而求得肌肉模型中所有节点的位移,从而实现以单视频驱动基于肌肉原理的人脸模型。采用此方法我们得到了逼真的表情和口型模拟动画。The paper presents a method to drive muscle-based facial model using single video. Compared with other conventional methods which drive facial model by using multi-view video sequences, this method first gets moving parameters of facial muscle by tracking specific control nodes on human face, then displacements of all nodes are gotten on facial model. In this way we can drive muscle-based facial model only using single video. We have gotten realistic human expressions and mouth shape by this method.

关 键 词：肌肉模型 人脸表情 MPEG4 单视频 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]