模式识别的经典算法，它是在1996年由Belhumeur引入模式识别和人工智能领域的。性鉴别分析的基本思想是将高维的模式样本投影到最佳鉴别矢量空间，以达到抽取分类信息和压缩特征空间维数的效果，投影后保证模式样本在新的子空间有最大的类间距离和最小的类内距离，即模式在该空间中有最佳的可分离性。因此，它是一种有效的特征抽取方法。使用这种方法能够使投影后模式样本的类间散布矩阵最大，并且同时类内散布矩阵最小。就是说，它能够保证投影后模式样本在新的空间中有最小的类内距离和最大的类间距离，即模式在该空间中有最佳的可分离性。-Classic pattern recognition algorithm, which is introduced in 1996 by Belhumeur pattern recognition and artificial intelligence. The basic idea of discriminant analysis of high-dimensional pattern is projected onto the sample optimal discriminant vector space, in order to achieve the classification of information extraction and compression feature space dimension effect, the projected sample assurance model in the new sub-space of the largest class the minimum distance and within-class distance, mode has the best separability in the space. Therefore, it is an effective method of feature extraction. Between the use of this method enables projection mode sample class scatter matrix largest and the smallest in the class scatter matrix simultaneously. That is, it can ensure that the samples have a minimum projection mode classes and maximum distance between the distance class, that has the best model separability in the space in the new space.