位置正则的支持向量域描述在人脸识别中的应用研究  

APPLICATION OF POSITIONAL REGULAR SUPPORT VECTOR DOMAINS IN FACE RECOGNITION

作　　者：熊昕[1] 曾青松[2] 

机构地区：[1]广州番禺职业技术学院教育技术与信息中心,广东广州511483 [2]广州番禺职业技术学院信息工程学院,广东广州511483

出　　处：《计算机应用与软件》2017年第5期163-167,共5页Computer Applications and Software

基　　金：广东省自然科学基金项目(2015A030313807)

摘　　要：支持向量域描述是一种有效的一分类数据描述方法,能够有效地对单一类别的数据进行表达,并能有效地降低负样本的干扰。应用支持向量域描述方法,将人脸图像集合投影到高维特征空间构建描述特征空间中人脸图像的超球体,并定义两个超球体之间的相似性度量,应用最近邻分类器进行分类。在基于集合的人脸识别应用标准数据库上测试了该方法,在Honda/UCSD、CMU Mobo和You Tube数据分别取得100%、97.55%和59.78%的识别率。实验结果表明,该方法是一种有效的基于图像集匹配的人脸识别方法。Support vector domain description is an effective method to describe a single class of data,and can effectively reduce the interference of negative samples. In this paper,the support vector domain description method is used to construct a hypersphere that describes the face image in the feature space by projecting the face image set into the high-dimensional feature space. And the similarity measure between two hyperspheres is defined and classified by nearest neighbor classifier. This method was tested on the standard database of face recognition based on collection. The recognition rate of Honda/UCSD,CMU Mobo and You Tube data were 100%,97. 55% and 59. 78% respectively.Experimental results show that the proposed method is an effective method for face recognition based on image set matching.

关 键 词：支持向量域描述 人脸识别 模式识别 集合匹配 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]