为进一步提高MPEG-4人脸动画的真实感,引入了独立分量分析,将其应用在MPEG-4模型基编码的人脸动画参数分析方面.通过对获取到的人脸动画参数训练数据应用独立分量分析。获得了一般人脸动画模式,调整独立分量分析空间参数可以生成特定人的任意表情.实验结果表明:基于独立分量分析空间的MPEG-4人脸动画具有较强的真实感和鲁棒性。
To gain the realistic MPEG-4 facial animation, Independent Component Analysis (ICA) is applied to the pattern extraction of facial animation parameters (FAPs). By studying the independent sources and their project on human expression, some steady independent components are extracted to come into being. Experiment results show that MPEG-4 facial animation based on ICA space is more robust and more realistic.