摘要
在人脸识别的某些应用中,最好能够找到原始特征的关键子集,减少不必要的特征计算和资源耗费,而不是得到所有原始特征的映射。主成分分析法(Principal Components Analysis,PCA)是目前比较常用的人脸识别算法,PCA将人脸图像映射到能很好地表征训练图像集的特征脸空间中,但是基于PCA的人脸识别的缺陷在于原始空间所有的特征都映射到了低维特征空间中,是基于最佳描述性特征子集。提出了一种新的基于PCA的特征选择方法,将特征选择与特征抽取相结合,对特征脸空间再进行特征选择,选择人脸原始特征集中最关键的特征,并将其应用在基于PCA的人脸识别中。
In some applications of face recognition,it might be more desirable to pick a subset of the original features than to find a mapping that uses all of the original features.The benefits of finding this subset of features lie in cost reduced computations and thus lower cost of sensors.Principal components analysis（PCA） is widely used in face feature extraction and recognition.The facial images are projected onto eigenfaces that best define the variation of the known test images.However,the PCA-based face recognition has the disadvantage that,on the basis of an optimal descriptive feature subset,measurements from all the original features are used in the projection to the lower dimensional space.Propose a new method for dimensionality reduction of a feature set by choosing a subset of original features that contains most of the essential information.This method,based on PCA,combines together feature selection and feature extraction.The proposed method has been successfully applied in choosing principal features in PCA-based face detection and recognition.
出处
《计算机技术与发展》
2011年第4期123-125,共3页
Computer Technology and Development
基金
江苏省自然科学基金(08KJB520008)
南京邮电大学人才引进启动基金(NY207137
NY207148)
关键词
人脸识别
PCA
特征脸
特征选择
face recognition
PCA
eigenface
feature selection
