摘要
以主成分分析和局部保持投影为理论基础,提出了一种同时考虑数据样本的全局和局部特性的大间距无监督正交特征提取算法,算法的目标函数采用大间距准则,避免了由于矩阵求逆带来的小样本问题,同时为了进一步增强算法的识别性能,对所求取的投影矩阵进行了正交化约束,最后人脸库上的实验结果表明所提方法的有效性.
A maximum margin unsupervised orthogonal feature extraction algorithm based on principal component analysis and locality preserving projection is proposed,in which both global and local features of the data samples are taken into account.The proposed method adopts the maximum margin criterion as object function and avoids the small sample size problem.To further enhance the recognition performance of the algorithm,orthogonal constraint projection matrix is given.Experimental results on face database demonstrate the effectiveness of the proposed method.
出处
《传感器与微系统》
CSCD
北大核心
2012年第4期143-145,共3页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(60975009
61170060)
安徽省自然科学基金资助项目(1208085QF123)
关键词
大间距
特征提取算法
目标函数
小样本问题
maximum margin
feature extraction algorithm
objective function
small sample size problem
