中文摘要：

针对面部识别问题提出了基于QR分解的模糊线性判别分析方法,并通过ORL、Yale和FERET人脸数据实验研究了该方法在不同距离下对面部识别率的影响;同时还研究了KNN分类器中K值的选择对面部识别率的影响。实验结果表明,距离的选取对面部识别率的结果有明显的影响。对不同的人脸数据集来说,KNN分类器中的K的选取也会对识别率有影响。对于ORL面部图像数据来说,在Minkowski距离下（m=3）,K=1时分类效果最好;对于YALE人脸数据,在Chebyshey距离下,K=5时分类效果最好;对于FERET人脸数据,在绝对距离下,K=1时分类效果最好。

英文摘要：

A fuzzy linear discriminant analysis method based on QR decomposition for face recognition problems is proposed.By means of experiments with ORL,Yale and FERET face databases,we study the affection of different distances in linear discriminant analysis method based on QR decomposition for face recognition rate.Furthermore,we also study the affection of different K-values in KNN classifier for face recognition rate.The experimental results show that the selection of distances has a significant impact for the results of the face recognition rate.For different face database,the K value of KNN classifier selection will also affect the recognition rate.For ORL face image data,in the Minkowski distance（m=3）,K=1 have the best classification results.For YALE face data,in the Chebyshey distance,K=5 have the best classification results.For the FERET face data,in the absolute distance,K=1 have the best classification results.