Abstract: In order to solve the problem of high dimensional image with complicated calculation, the paper proposes a method combined with weighted wavelet analysis and DCT for face recognition. The method decomposes face image by wavelet, extracts the DCT transform coefficients of low-frequency and weighted high-frequency as feature vectors. It uses the weighted distance for classification. Compared with the traditional PCA algorithm as well as wavelet combining PCA algorithm, the experimental results on ORL and YALE face databases show that the proposed algorithm can improve the training time and the recognition rate.

Key words: face recognition, weighted wavelet analysis, Discrete Cosine Transform（DCT）, weighted distance

摘要： 针对图像维数过高，计算复杂的问题，提出一种基于加权小波分析和DCT的人脸识别方法，通过对人脸图像进行小波分解，提取低频和加权高频分量的DCT变换系数作为识别特征向量，采用加权距离进行分类识别。该方法在ORL和YALE人脸库上进行了测试比较，结果表明，无论训练时间还是识别率，都优于传统的PCA方法，和小波结合PCA的方法相比较，识别率也明显提高。

关键词: 人脸识别, 加权小波分析, 离散余弦变换（DCT）, 加权距离