摘要
为了解决人脸识别中的光照干扰问题,提出基于最近正交矩阵的二维鉴别投影方法.首先利用奇异值分解获取人脸图像矩阵的最近正交矩阵表示;然后利用获得的最近正交矩阵分别构建基于最近正交矩阵的类内散度和基于最近正交矩阵的类间散度;最后通过最大化类间散度同时最小化类内散度获取二维鉴别投影,并通过该投影获得低维特征.在Yale,CMU-PIE以及AR数据库上进行人脸识别实验,证明了该方法的有效性.
To solve the interferential problem of illumination in face recognition,this paper proposes a methodcalled two dimensional discriminative projection based on nearest orthogonal matrix(2DDP-NOM).Itobtains nearest orthogonal matrix representation of face image matrix by singular value decompositionfirstly;then constructs the intra-class scatter based on nearest orthogonal matrix and the inter-class scatterbased on nearest orthogonal matrix by nearest orthogonal matrix;lastly,obtains two dimensional discriminativeprojection by maximizing the inter-class scatter and minimizing the intra-class scatter simultaneouslyand gets the low dimensional features by this projection.Experiments are performed on Yale,CMU-PIE andAR databases and the experimental results demonstrate the effectiveness of2DDP-NOM.
作者
殷俊
孙仕亮
Yin Jun;Sun Shiliang(College of Information Engineering, Shanghai Maritime University, Shanghai 201306;Department of Computer Science and Technology, East China Normal University, Shanghai 200241;Key Laboratory of Intelligent Perception and Systems for High-dimensional Information of Ministry of Education (Nanjing University of Science and Technology), Nanjing 210094)
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2017年第8期1457-1464,共8页
Journal of Computer-Aided Design & Computer Graphics
基金
国家自然科学基金(61370175,61603243)
上海市自然科学基金(13ZR1455600)
高维信息智能感知与系统教育部重点实验室创新基金(JYB201607)
关键词
最近正交矩阵
维数约简
鉴别投影
人脸识别
nearest orthogonal matrix
dimension reduction
discriminative projection
face recognition