基于2DPCA和边界训练样本的人脸识别方法
首发时间:2009-11-12
摘要:提出了一种基于二维PCA(2DPCA)和边界训练样本的人脸识别方法.某类的边界样本作为该类的特殊样本,在特征空间的投影值更有可能构成该类投影区域的边界,因此,利用边界样本训练分类器,有利于提高其识别率。基于此,该识别方法选择边界样本作为训练样本,在特征提取方面采用2DPCA算法,2DPCA算法直接利用原始图像矩阵构造协方差矩阵,降低了原始特征的维数,特征抽取更直接。基于ORL和Yale人脸数据库的实验结果表明,该样本选择方法提高了分类器的识别性能;同时也表明,在人脸识别方面,2DPCA比PCA具有更高的识别率,且特征提取更有效,速率也更快。
关键词: 2DPCA 边界样本 人脸识别 特征提取 分类识别
For information in English, please click here
Face Recognition Based on 2DPCA and Boundary Training Samples
Abstract:A face recognition method based on two-dimensional PCA (2DPCA) and boundary training samples presents. Boundary samples as special samples in a class, its projective values in feature space are more probably composed the boundary of projective area. It’s helpful to increase recognition rate of classifier while using boundary samples to train it. So, we choose the boundary samples as the training samples, and 2DPCA algorithm in feature extraction. By way of 2DPCA algorithm, the original image matrix is directly utilized to construct covariance matrix, so, the dimension of original pattern reduces and discriminant features gains directly. The experimental results on ORL and Yale face database demonstrate that this method of choosing training samples is efficient; at the same time, they demonstrate that recognition rate of 2DPCA in face recognition is higher than that of PCA; and that 2DPCA is more valid and rapid than PCA in feature extraction.
Keywords: 2DPCA boundary samples face recognition feature extraction classification and identification
基金:
论文图表:
引用
No.3664450161212580****
同行评议
共计0人参与
勘误表
基于2DPCA和边界训练样本的人脸识别方法
评论
全部评论0/1000