基于局部参数运动模型的人脸表情特征提取

Facial Expression Features Extraction Based on Local Parameterized Models

提出了一种基于局部参数运动模型提取人脸表情运动参数的方法.对图像序列进行预处理,提取出能够体现表情的五官特征;用块匹配法跟踪了特征区域的关键点;基于特征点匹配结果和局部参数运动模型,建立特征区域的运动方程;通过非线性约束优化方法估计局部运动参数.实验结果表明,该方法可行有效.

An approach to extracting motion parameters of facial expression based on local parameterized models is proposed in this paper. First, a set of image sequence is preprocessed, and the facial features which can reflect expression are extracted. Secondly, the method of block-match is used to track the critical feature points. Then, the affine motion equation based on the result of block-match and the local parameterized model is created. Finally, a nonlinear constraint optimization way is adopted to estimate the motion parameters. The experiment result demonstrates that the method is feasible.