基于机构学和肌肉模型的单视频驱动人脸情感表达仿真  

Drive facial animation by using single video based on the mechanism theory and muscle model

作　　者：张抗[1] 樊养余[1] 吕国云[1] 马元媛[1] 王毅[1] 

机构地区：[1]西北工业大学电子信息学院,西安710072

出　　处：《中国图象图形学报》2011年第3期449-453,共5页Journal of Image and Graphics

基　　金：国家高技术研究发展计划(863)项目(2007AA01Z324)

摘　　要：描述一种3维人脸模型的单视频驱动方法。该方法在传统肌肉模型的基础上,根据嘴部运动特性采用机构学原理建立嘴部运动控制模型,通过视频图像序列跟踪得到特征点的运动规律曲线,进而驱动眼部、嘴部及面部其他部分的网格点运动,产生具有真实感的面部表情动作。仿真结果表明,采用此方法可以得到逼真的人脸表情模拟动画。A method to drive 3D facial model using single video was presented. Based on the movement characteristics, the mouth was controlled by using a mechanism model. The motion trace of the feature points was acquired by tracking video image sequence. Then we drive the grid nodes on the facial model to obtain a realistic facial animation. The results show that this method could simulate the facial movements and expressions very well.

关 键 词：机构学 肌肉模型 情感表达 人脸仿真 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]