甘俊英,李高尚
摘 要:利用离散小波变换对人脸图像进行压缩,提取人脸的低频分量,有效去除人脸图像高频分量的影响;再利用二维主元分析对小波变换后的人脸低频分量实行提取特征;然后使用核主元分析再次提取特征;最后用最小距离分类器完成人脸识别。基于ORL人脸数据库的实验结果表明,该算法能提高人脸识别率,有效减少计算量和降低计算复杂度。
关键词:小波变换;2DPCA算法;KPCA算法;人脸识别
中图分类号:TP391.41文献标识码:B
文章编号:1004-373X(2009)20-051-03
Face Recognition Based on DWT,2DPCA and KPCA
GAN Junying,LI Gaoshang
(School of Information,Wuyi University,Jiangmen,529020,China)
Abstract:In this paper,human face image is compressed and low-frequency component is extracted by way of Discrete Wavelet Transform (DWT).In this way,the influence of high-frequency component of human face image is discarded effectively.Then the features of low-frequency component are extracted by way of two dimensional principal component analysis (2DPCA).In the mean time,Kernel Principal Component Analysis (KPCA) is applied in feature extraction.Finally face recognition is performed by the Nearest Neighbor classifier.Experimental results on Olivetti Research Laboratory(ORL)face database show that face recognition rate is increased,amount of computation and the complexity are reduced.
Keywords:wavelet transform;2DPCA algorithm;KPCA algorithm;face recognition
0 引 言
近年来,人脸识别技术得到了很大发展,许多优秀的方法和算法相继被提出[1,2]。小波变换(Wavelet Transform,WT)是一种基于多层次函数分解的数学工具[3-5]。人脸图像经过小波变换后,可分解成近似分量与细节分量。近似分量代表信号的低频分量,细节分量代表信号的高频分量。近似分量保留图像的基本信息,噪声分量或变化比较快的分量,其主要能量一般集中在细节分量中。Nastar等研究了人脸外观变化与频谱变化之间的关系,指出人脸的光照、旋转扭曲和面部表情只影响图像中的高频分量,人脸图像的低频分量仍然保持稳定。因此,采用小波变换后仅保留近似分量,可有效克服光照、姿态和表情变化对人脸识别率的影响。
2004年Jian Yang等人提出二维主元分析(Two-dimensional Principal Component Analysis,2DPCA),并将该算法应用于人脸识别[6]。2DPCA算法是以人脸样本的二维灰度图像矩阵计算样本的总体离散度的,所以它有效降低了待识别图像的维数。Jian Yang等的实验结果表明,2DPCA算法的识别率和计算时间均优于主元分析(Principal Component Analysis,PCA)算法。……