为了提高计算机合成人脸表情动画的后期制作效率,提出一种基于时空的人脸表情动画编辑方法.首先使用基于拉普拉斯的网格变形技术将用户的编辑效果在空间域传播到整个人脸模型,很好地保留了中性人脸模型的几何细节特征,从而提高合成表情的真实感;然后使用高斯函数将用户编辑效果在时间域传播到邻近表情序列,保持人脸表情动画的平滑过渡,所合成人脸表情动画与初始给定数据保持一致.该方法为用户提供了人脸表情动画编辑的局部控制,可由用户指定编辑的影响范围,使得编辑效果在指定范围内自然传播.实验结果表明,文中方法所合成人脸表情动画自然、真实,有效地提高了数据驱动人脸表情动画的编辑效率.
Abstract: To improve the post-editing efficiency of the synthesized facial animation, this paper proposes a spacetime-based method to edit the facial animation. It propagates the users' editing among the input facial model with the Laplacian-based mesh deformation, which can preserve the details of the initial facial model and guarantee the reality of the synthesized expression. Then, a Gaussian function is employed to propagate the editing effect to the surrounding frames in a fast way. The approach retains the high consistency between the edited facial animation and the original facial animation during the propagation. It provides users with the local control of the facial animation editing. Users can assign the influence range of their editing and the editing result is propagated to the surrounding frames naturally. Experimental results show that our approach can generate natural and realistic facial animations. It can efficiently improve the editing efficiency for the data-driven facial animation.