摘要
为了提高人脸识别算法的识别率,提出了一种基于局部奇异值分解(Local Singular Value Decomposition,LSVD)和监督拉普拉斯特征映射(Supervised Laplacian Eigenmap,SLE)的人脸图像识别方法。由于奇异值向量具有良好的稳定性、转置不变性等特点,首先利用局部奇异值分解方法从人脸图像中提取特征向量;然后采用监督拉普拉斯特征映射算法对已获取的人脸特征进行维数约简。在Yale和ORL人脸库上的实验结果表明,该算法能有效地提高人脸识别的性能。
Because of the characteristics such as good stability and transpose invariance of singular value vector, feature vectors are extracted from face image by using local singular value decomposition method firstly. Then dimensionality reduction is applied on facial features which have been acquired by supervised laplacian eigenmap algorithm. Experimental results on Yale and ORL face database show that the proposed algorithm can effectively improve the performance of face recognition.
作者
沈杰
嵇春梅
王正群
王明辉
钱亚芹
SHEN Jie JI Chunmei WANG Zhengqun WANG Minghui QIAN Yaqin(Modern Educational Technology Center, Yancheng Institute of Technology, Yancheng Jiangsu 224051, China College of Mechanical and Electrical Engineering, Yancheng Vocational Institute of Industry Technology, Yancheng Jiangsu 224005, China College of Information Engineering, Yangzhou University, Yangzhou Jiangsu 225009, China)
出处
《盐城工学院学报（自然科学版）》
CAS
2016年第4期16-20,共5页
Journal of Yancheng Institute of Technology:Natural Science Edition
基金
国家自然科学基金面上项目(61402395)
江苏省文化科研项目(15YB38)
江苏省现代教育技术研究课题(47937)
关键词
人脸识别
流形学习
局部奇异值分解
监督拉普拉斯特征映射
face recognition
manifold learning
local singular value decomposition
supervised laplacian eigenmap
