基于小波变换和二维非负矩阵分解的人脸识别算法

摘要

This paper advanced a new fusion recognition algorithm 2DDWT +2DNMF through synthesizing two advantages of 2DDWT and 2DNMF. It firstly used wavelet transform to decompose face image to four sub-band images. Afterwards it focused on fusion of three high frequency sub-band images, and then used 2DNMF algorithm to extract the features of low frequency sub-band image and fusion image. At last it weighted these extracted features. Experimental results from ORL and YALE face image database show that compared with PCA,SVD,NMF and 2DDWT + NMF algorithms,the new fusion method can shorten training time and improve recognition rates effectively.%结合二维离散小波变换(2DDWT)和二维非负矩阵分解(2DNMF)两者的优点,提出了一种新的人脸识别融合算法2DDWT+2DNMF.首先利用小波变换把人脸图像分解成四个子块频带区域,并对三个高频子块进行图像融合,然后对低频子块和融合图像进行二维非负矩阵分解以提取特征,进而对特征数据进行加权处理.ORL和YALE人脸数据库中的识别实验表明,与PCA、SVD、NMF以及2DDWT+ NMF算法相比,新融合算法能有效缩短训练时间和提高识别率.