多民族人脸表情变换

Multinational Country Face Expression Changes

我国是一个多民族国家。不同民族在表达相同表情时,会因文化,地域和习惯的不同,产生一定的区别。为了表达多民族人脸表情变换的特点,实现了一种以民族人脸中性表情为输入,依据各民族人脸表情特点,自动生成的民族人脸表情变换方法。首先利用构建的多民族人脸表情数据库标准人脸库,多民族人脸表情库,以及多民族人脸表情AU部件库,统计多民族人脸表情变化时的规律,利用主动形状模型ASM对民族人脸进行特征点定位。其次,根据统计不同民族人脸的表情,获得民族表情变换参数,结合一般人脸表情变换参数计算民族人脸表情变换特征点的形变位置。最后,使用纹理映射和仿射变换生成变换表情后的人脸图像。实验结果显示,所提方法能够仿真出不同民族人脸表情变换的基本表情。

China is a multi-ethnic country. In order to express the multi-ethnic facial expression characteristics, a method generated ethnic facial expression automatically is presented aimed at national neutral facial expressions photo according to the characteristics of each ethnic facial expression. Firstly, the changing rule statistics is gathered by constructed Multi-ethnic Facial Expression Standard Database(MFESD), Multi-ethnic Facial Expression Database(MFED), and Multi-ethnic Facial Expression AU Database(MFEAUD) and the facial characteristic points are located by ASM method. Secondly, the deforming position of the multi-ethnic facial expression characteristic points are calculated combining the multi-ethnic facial expression parameter gathered by the statistics of different ethnic facial expression with the popular parameter of facial expression transform. Finally, the transform image is generated by texture mapping and affine transformation. The result of experiment shows that the method proposed could simulate multi-ethnic facial base expression.