基于三维人脸建模的多姿态人脸识别

Face recognition with multiple pose based on 3-D face model

待匹配人脸图像与原始图像存在姿态和光照的差异,是自动人脸识别的两个主要瓶颈问题。给出了采用三维人脸模型来解决人脸姿态的变化对人脸识别的影响问题。通过正侧面图像,利用B样条曲线与径向基函数相结合的方法进行三维人脸重建,生成三维人脸模型库。分别计算待匹配人脸图像的3个自由度,快速估计出人脸的姿态;结合待匹配人脸图像的姿态参数及三维人脸模型库,获得与待匹配图像相同姿态的三维人脸模型库中的二维人脸图像。最后完成了相同人脸姿态的二维人脸识别。实验结果证明,该方法无需复杂的设备、简单易行、识别时间短,是一种非常实用的解决人脸姿态问题的识别方法。

The differences between matching face image and original face image in pose and illumination are bottleneck problems in automatic face recognition. 3-D face model is used to tackle pose problems in face recognition. Main processes are as follows： at first, RBFs （radial basis functions） and B-spline curves methods are combined to process frontal and profile images in order to carry on 3-D face reconstruction and to construct a 3-D face model database. Then face pose parameters is estimated rapidly when three freedom degrees of face image are calculated respectively. Later, the pose parameters of matching face image and the 3-D face model are used to produce a new face image which possesses the same pose parameters as the matching face image. Finally, the 2-D face recognition is done with the same face pose. Experiments prove that the method is effective in tackling face pose problems and it is simply and quickly operated.