人脸图像像素较高，不利于后续识别，里面有一些维数约简的算法，浙大蔡登写的-Face image pixels high, is not conducive to the subsequent recognition, there are some dimensionality reduction algorithms, Zhejiang written by Cai Deng