说明：  1.首先读入10副人脸图像，这些图像大小相等，是一个人的各种表情图片。 2.把图片转为1*pq的一维形式，p是图像宽，q是图像高。这时我们的S矩阵就是10行，每行是pq维的向量。 3.然后我们在S上执行PCA算法，设置K=5，求得5个特征向量，这5个特征向量就是我们求得的特征脸，用这5个特征脸图像，可以近似表示之前的十副图像。
(1. First read in 10 face images, which are equal in size and are all kinds of expression pictures of a person. 2. Turn the picture into a one-dimensional form of 1 * p q. P is the image width and Q is the image height. At this time, our S matrix is 10 lines, and each line is the vector of PQ dimension. 3. Then we execute PCA algorithm on S, set K = 5, and get five eigenvectors, which are the eigenfaces we get. With these five eigenfaces, we can approximate ten previous images.)