摘要
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.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第7期1728-1731,共4页
Computer Engineering and Design
基金
广东省自然科学基金项目(07010869
032356)
广东省江门市科技攻关基金项目(江财企[2006]59号)
关键词
人脸识别
三维人脸重建
姿态参数
径向基函数
B样条曲线
face recognition
3-D face reconstruction
pose parameter
radial basis functions
B-spline curves
