基于轮廓线曲率特征的人脸分类及识别

Curvature\|feature\|based method for human face classification and recognition

文中提出一种人脸分类及识别方法。在对人脸图像进行预处理之后，提取出人脸的轮廓，再细化，去除边缘噪声，得到人脸的轮廓线。对此轮廓线逐点计算曲率得到曲率线。将不同的曲率线之间的互相关函数定义为曲率线之间的相似性，并由此对人脸进行分类和识别。由于曲率几何特征明显，具有位移和旋转不变性，所提出的方法有较好的鲁棒性，并且得到了良好的分类效果。

A method for human face classification and recognition is proposed in this paper. Firstly, the human face image is pre\|processed and the edge is extracted, and then the method is used to get a continuous edge. Finally, the continuous human face contour curve is obtained by thinning and removing the edge noise. After smoothing the curve, the curvature of every point of the contour is calculated. Then a curvature curve is obtained which represents a human face. Since curvature feature is shift invariant and rotation invariant, it is more robust than the human face contour. The correlation of every two curvature curves is applied to denote the similarity of the two different curves. It is proved that the method can result in good classification and recognition effects. \;