Thirdly, some eigenvectors are extracted for training and predicting face images by casting singular value decomposition (SVD) on face data.

• 第三,对人脸区域数据进行奇异值分解,选取若干特征向量用来训练及判别。