基于颜色信息和几何不变性的人脸检测定位

由先验知识我们知道,2D人脸正面图像几何对称;然而,当姿态发生变化时,对于人脸这样的不规则3D几何体,不同的视角、不同的摄像机参数使得在透视成像下得到的图像也不同,并且发现正面人脸具有的对称特性也消失了,因此3D人脸的识别是十分困难的;提出一种从人脸特征的结构特殊性出发,利用2D人脸形状、面部特征等内在的几何约束关系构造射影不变的特征参数、特征关系的射影不变性,同时结合颜色物理信息的人脸检测定位方法,有效地避免了构造3D人脸模型的难题,增强了实验结果的效率、可靠性和稳定性.

Face Location Based on Color And Geometric Invariance

From comman knowPedge we know 2D human faces are eudipleural. But in many case, the face symmetric disappear . Now the 3D face recognition is a difficult problem because of the effects of viewpoint and extrinsic parameters of camera, the obtain image obtained from a same face is different. The paper advance a method based on 2D facial character geometry invariance through skin color segmentation, according to the facial feature geometric relationship is invariance under Projective transformation. The method can effectively eliminate the difficult of the construct of 3D face model. The satisfactory experiment results prove that the algorithm is valid.