基于Gabor变换的增强2DPCA人脸识别算法
首发时间:2011-03-08
摘要:本文提出一种基于二维Gabor小波特征矩阵的增强2DPCA人脸识别算法。该算法利用了Gabor小波变换对位置和光线的不敏感性及良好的纹理提取特性,以二维Gabor小波矩阵描述人脸特征。使用增强2DPCA方法进行特征降维,先对二维Gabor小波矩阵做行方向的2DPCA变换,再对生成的特征矩阵进行列方向的2DPCA变换。分别在ORL人脸库和Yale人脸库采用最近邻分类器进行试验,结果表明该方法性能优良,具有较高的识别率。
关键词: 模式识别 人脸识别 特征提取 Gabor小波 二维主成分分析(2DPCA)
For information in English, please click here
Novel enhanced 2DPCA face recognition algorithm based on 2D Gabor transform
Abstract:This paper presents an enhanced 2DPCA face recognition algorithm based on 2D Gabor wavelet feature. This algorithm makes use of attributes of Gabor filter, which is insensitivity to position and light. In addition, the Gabor filter has fine characteristics of good texture extraction. Therefore, this algorithm uses 2D Gabor Wavelet feature to describe features on human face. Then, the enhanced 2DPCA algorithm is used to compress the Gabor Wavelet feature matrix. Firstly, the method uses 2DPCA to compress the Gabor Wavelet feature matrix in the row direction. Secondly, the method uses 2DPCA to compress the new matrix in the column direction. Nearest Neighbor (NN) is used to construct classifier and classify graphs of face. The experimental results based on ORL and Yale human face databases show that this method has excellent performances.
Keywords: Pattern Recognition Face Recognition Feature Extraction Gabor wavelet (GW) Two-dimensional principal component analysis (2DPCA)
基金:
论文图表:
引用
No.4413433494337129****
同行评议
共计0人参与
勘误表
基于Gabor变换的增强2DPCA人脸识别算法
评论
全部评论0/1000