改进的线性判别分析及人脸识别

为有效解决传统LDA（线性鉴别分析）的小样本规模问题，提出一种改进的LDA算法。首先对样本进行无损降维；然后在Fisher准则基础上，用散度矩阵差代替散度矩阵的比值，避免对类内矩阵求逆的同时也降低了计算复杂度，实现有效的特征抽取；最后实现对人脸的识别。实验结果表明，该算法是有效的，优于传统LDA方法。

A Modified Linear Discriminant Analysis and Face Recognition

For effectively solve linear discriminant analysis in small sample problem.This paper presented an improved linear discriminant analysis(LDA) algorithm for face recognition.First,the method carries on the lossless dimensionality reduction to the sample.Then in the Fisher criterion foundation,it replaces the divergence matrix with the divergence matrix difference ratio,to avoid the kind of matrix inverse also reduce the complexity of the calculation,the effective realization of the feature extraction.Finally...