关键点匹配三维人脸识别方法*

提出了一种新颖的三维人脸识别算法，其基本思路是，把代表人脸的三维点云沿X、Y或Z轴旋转，反复多次把3D人脸关键点投影到2.5D图像上，然后提取2.5D图像的关键点并进行标记，而用这些比原来小得多的关键点代替原来的面扫描。面对未知的待测人脸首先通过执行相同的多视角特征点提取技术提取关键点，然后应用一个新的加权特征点匹配算法进行识别。通过用GavabDB三维面部识别数据集进行试验评估，这个方法对中性表情人脸可获得高达94％的识别精度，对人脸表情辨识（如微笑）的准确率也超过了88％。实验结果表明，此方法在识别精

Method of 3D face recognition based on keypoint matching

This paper proposed a novel algorithm for 3D face recognition based on keypoint matching. Its idea was to rotate each 3D point cloud representing a face around the x, y or z axes, iteratively projecting the 3D points onto 2.5D images. It extracted the keypoints from 2.5D images, set of keypoints replaced the original face scan, performed test faces the same keypoint extraction technique, and secondly using a new weighted keypoint matching algorithm to recognize face. Evaluation using the GavabDB 3D face recognition dataset, the method achieved up to 94% recognition accuracy for faces with neutral expressions, and 88% accuracy for face recognition with expressions (such as a smile).The experiment results show that this method gets remarkable progress in recognizing accuracy.