摘要
为了生成真实感的三维人脸动画,提出了一种新的兼容于MPEG-4的人脸动画方法.采用基于均匀网格重采样和网格简化相结合的方法对齐原型三维人脸,并在三维模型上定位FDP和FAP控制点,建立兼容于MPEG-4的人脸动画模型,根据已知的人脸动画定义表构建适合于该方法提出的三维模型的人脸动画定义表.实验结果表明,该人脸动画方法可根据一幅给定的人脸图像自动重建特定人的三维人脸,并根据FAP参数序列驱动模型,生成真实感较好的人脸动画图像.
A novel 3-D facial animation compatible with MPG-4 is proposed based on morphable model. Uni- form mesh resampling combined with mesh simplification based algorithm is adopted to compute correspon- dence between prototypical 3-D faces. Having located FDP and FAP points on 3-D model, 3-D facial anima- tion model compatible with MPEG-4 is constructed. Facial animation definition table corresponding to the 3-D facial animation model is constructed according to a known one. Experimental results show that given a face image, specific 3-D face of the person can be constructed automatically by the animation model, and vivid facial animation images can be generated according to FAP values.
出处
《北京工业大学学报》
EI
CAS
CSCD
北大核心
2007年第10期1102-1106,共5页
Journal of Beijing University of Technology
基金
国家自然科学基金(60375007
60533030)
北京市自然科学基金(4061001).