主分量分析方法，用一组维数为数不多的特征尽可能精确地表示样本的特征，通常采用训练样本总体协方差矩阵的特征向量系作为展开基(即K-L坐标轴)-Principal component analysis, said the characteristics of the samples as accurately as possible using one of the few characteristics of a group of dimension, usually the overall training sample covariance matrix of the Eigenvector as expand the base (ie, the KL axis)