摘要
人脸识别因其高效、安全和非接触性的特点,在公共信息安全领域得到了广泛应用.针对传统主元分析方法(PCA)和随机主元分析法(Random PCA)在实际应用中存在抗干扰能力差、识别率不高以及2种方法特征融合后计算复杂的问题,提出了一种基于随机主成分分析+粗糙集(Random PCA+rough set)的人脸识别方法.该方法用PCA提取人脸的全局特征,用Random PCA提取人脸图像的局部特征,再将这2种特征通过串联的方式构建特征子空间.在特征子空间里用粗糙集去提取最具区分度的特征,从而有效减少了分类时的计算复杂度并提高了识别率.实验结果表明:该方法较传统PCA方法的识别率和识别时间分别提高了7.09%和6.06%.
Face recognition has been widely used in the field of public information security due to it＇s high efficiency, safety, and non-contact. The problems of traditional methods principal component analysis （PCA） and Random principal component analysis（ Random PCA）is lack of anti-interference and low recognition rate, otherwise the problem of in fusion of PCA and Random PCA is the calculate time is too long. In order to solve these problems, a method based on Random PCA plus rough set is proposed for face recognition. The method first exploit PCA and Random PCA extract the global feature and local feature, respectively. And then cascade the global feature and the local feature to construct the feature subspace. At last exploit rough set to extract the most distinguish feature from the feature subspace , therefore the method can improve the ability of anti-interference and recognition rate. Compared with the traditional method PCA, the results show that the recognition rate and recognition time improved 7.09% and 6.06%, respectively.
出处
《江西师范大学学报（自然科学版）》
CAS
北大核心
2016年第5期487-491,共5页
Journal of Jiangxi Normal University(Natural Science Edition)
基金
江西省科技计划(20133BBE50035)资助项目
关键词
人脸识别
主元分析法
粗糙集
特征子空间
face recognition
principal component analysis
rough set
feature subspace
