融合小波与2DPCA的SVM人脸识别

近年来,人脸识别由于其诱人的应用前景再次成为模式识别领域的研究热点。分析了小波变换、2DPCA以及SVM 3种方法在人脸识别中各自的优势,提出了融合小波和2DPCA进行SVM人脸识别的方法。首先对原始图像采用小波分解提取低频信息,忽略高频分量;然后利用2DPCA进行特征提取;最后把降维后的数据输入SVM进行分类识别。该方法在ORL、Yale人脸库上的实验表明,与传统的方法相比,不但可以提高识别率,而且所用时间明显减少。

SVM Face Recognition Using Wavelet Transform and 2DPCA

In recent years,face recognition becomes a hot field of pattern recognition once again because of its attractive prospect.The respective advantages of wavelet transform,2 DPCA and SVM is analyzed in face recognition,and a SVM approach to face recognition is proposed based on wavelet transform and 2 DPCA.Firstly,the original images are decomposed into low frequency images by applying wavelet transform and ignored the highfrequency components.Then 2DPCA is used to deal with feature extraction.At last,SVM is made use of the feature to do classification.Its efficiency and superiority are clarified by comparative experiments on ORL and Yale face data.