摘要
本文提出了一种称为M2PCA+FDA的新的人脸识别方法.新方法从模式的原始数字图像出发,先对样本图像进行分块,对分块得到的子图像矩阵采用PCA进行特征抽取,从而得到能代替原始模式的低维的新模式,然后,对新模式施行“Fisher-faces”方法,实现模式的分类.其特点是能有效地抽取图像的局部特征,正是这些特征使此类模式区别于彼类.在ORL和NUST603两个人脸数据库上对M2PCA+FDA方法进行了测试,实验的结果表明,本文提出的方法在识别性能上优于“Fisher-faces”方法和PCA方法.
In this paper, a new technique called M2PCA＋FDA is developed for human face recognition. First, in proposed approach, the original sample images are divided into smaller modular images, which are also called sub-images, then, for feature extraction, the well-known PCA method can be directly used to the sub-images obtained from the previous step, and the new lower dimensionality patterns that can replace the original patterns are obtained. In the end, the classical Fisherfaces method is performed on the reductions for the pattern classification. The advantage of the represented way when compared with conventional PCA method on original images is that the local discriminant features of the original patterns can be efficiently extracted by the modular PCA, which are available to differentiate one class from another. To test M2PCA＋FDA and to evaluate its performance, a series of experiments will be performed on two human face image databases： ORL and NJUST603 human face databases. The experimental results indicate that the performance of the new method is obviously superior to that of both Fisherfaces and PCA.
出处
《小型微型计算机系统》
CSCD
北大核心
2006年第10期1943-1947,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60472060)资助
江苏省自然科学基金项目(05KJD520036
04KJD520037)资助.
关键词
线性鉴别分析
主成分分析
特征抽取
分块PCA
人脸识别
linear discriminant analysis （LDA）
principal component analysis （PCA）
feature extraction
modular principal component analysis（Modular PCA）
face recognition
