说明：   在特征提取阶段，研究了PCA, 2DPCA, (2D) 2PCA, DiagPCA, DiagPCA-F-2DPCA等多 种方法。不同于基于图象向量的PCA特征提取，由于2DPCA, (2D) ZPCA, DiagPCA和 DiagPCA-I-2DPCA的特征提取都直接基于图象矩阵，计算量小，所以特征的提取速度明 显高于PCA方法。
(In the feature extraction stage, the study of the PCA, 2DPCA, (2D) 2PCA, DiagPCA, DiagPCA-F-2DPCA and other methods. Vector is different from the PCA-based image feature extraction, as 2DPCA, (2D) ZPCA, DiagPCA and DiagPCA-I-2DPCA the feature extraction are directly based on image matrix, a small amount of calculation, so the speed of feature extraction method was significantly higher than PCA .)