结合子空间投影和流形学习的人脸识别系统研发  

Research and Design of Face Recognition System Combined with Subspace Projection and Manifold Learning

作　　者：邓素娟[1] 周鑫燚[2] 唐年庆[3] 

机构地区：[1]四川省内江市第一中学,四川内江641112 [2]成都师范学院计算机科学系,成都611130 [3]内江师范学院计算机科学学院,四川内江641112

出　　处：《成都师范学院学报》2015年第3期109-112,共4页Journal of Chengdu Normal University

摘　　要：人脸识别是一种广泛使用的生物特征识别技术。文章提出一种结合子空间投影和流形学习的高效人脸识别方法。该方法利用二维线性鉴别分析提取人脸的不变性特征,能有效提取人脸图像的局部特征并用于人脸识别。实验结果显示,此方法能有效克服光照、表情和遮挡等情况,具有较高的识别准确率。Face recognition is a widely used technology in the feature recognition on biology. This paper proposes an improved method for face recognition by applying 2 - D linear discriminant analysis that combines subspace projection and manifold learning. The proposed method is able to effectively extract the local features from facial images for face recogni- tion. The resuhs of experiments show that this method can accurately recognize facial images, even with lluminalion, expression and occluder.

关 键 词：人脸识别 子空间投影 流形 二维线性鉴别分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]