基于自适应阈值PCA的多目标人脸识别方法研究  

Research on multi-target face recognition based on adaptive threshold PCA

作　　者：郝刚[1] 梁鹏[1] Hao Gang;Liang Peng(School of Computer Science,Guangdong Polytechnic Normal University,Guangzhou 510665,China)

机构地区：[1]广东技术师范学院计算机科学学院,广东广州510665

出　　处：《信息技术与网络安全》2018年第9期49-51,共3页Information Technology and Network Security

基　　金：广东省科技厅协同创新与平台环境建设项目(2017A040405058);广东省自然科学基金博士启动项目(2015A030310340);广东省大学生攀登计划重点项目(pdjh2017a0292)

摘　　要：为有效解决人脸识别中多类分类问题,提出一种基于自适应阈值PCA的多目标人脸识别方法,该方法基于PCA原理,将多目标分类问题转化为多个二分类问题,利用ROC曲线确定单个二分类器的最佳阈值,测试样本的识别结果由所有二分类器投票产生。在ORL人脸库上的实验表明,与传统PCA的多目标人脸识别方法相比较,本文算法的识别率可提升3. 3%左右,汉明损失可降低0. 007 9左右。In order to effectively solve the multi-class classification problem in face recognition,a multi-objective face recognition method based on adaptive threshold PCA is proposed. The method is based on the principle of PCA and transforms the multi-objective classification problem into multiple two-classification problems. The optimal threshold of a single binary classifier is determined by ROC curve,and the recognition results of the test samples are generated by voting by all the two classifiers. The experiments on ORL face database show that compared with the traditional PCA method,the True Positive Rate of proposed algorithm can be increased by about 3. 3%,and the Hamming Loss can be reduced by about 0. 007 9.

关 键 词：人脸识别 ROC曲线 主成分分析 多类分类问题 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]