中文摘要：

主成分分析（PCA）是一种经典算法,可用于人脸识别系统。它基于降维的方法提取样本人脸图像中的主要成分,并将待识别的人脸图像映射到训练集中,经比对后得出识别结果。但在此基本方法中光照变化是影响判别结果的一个重要因素。为克服此问题,在此提出一种新方法,即首先基于中值思想得出较局部二值模式改进的灰度图像,然后借助主成分分析思想去除一些冗余特征,并且再次用PCA算法对图像进行识别。

英文摘要：

The principal component analysis（PCA）is a classical algorithm which can be used for the face recognition sys tem.It extracts the main component in the sample face images by reducing the dimensionality,maps the face images which need to be identified to the training set,and then figures out the identification outcome by comparison.However,the illumination con dition is a very important factor affecting the result of the discrimination.To solve this problem,a new method is proposed,in which gray images that are better than LBP are obtained first on the basis of the idea of intermediate values,some redundant fea tures are removed by the aid of principal component analysis,and then the images are identified again by the PCA algorithm. The experimental results demonstrated the effectiveness of this method.