直接正交鉴别保局投影算法

针对保局投影(LPP)及其衍生出的算法在人脸识别时须先采用主成分分析(PCA)算法对高维样本降维后才能应用,本文基于正交鉴别保局投影(ODLPP,orthogonal discriminal locality pre-serving projection)算法,提出了一种直接ODLPP(DODLPP)算法,利用拉普拉斯矩阵性质进行了相应的矩阵分解,可直接从高维样本的原始空间中提取投影矩阵。为解决ODLPP算法的小样本问题,给出先求解局部类内散度矩阵的零空间,然后再最大化类间散度矩阵的求解思路。人脸库上的实验结果表明所提算法的有效性。

Direct orthogonal discriminant locality preserving projections method

A series of feature extraction algorithms based on locality preserving projection have been proposed.PCA algorithm must be firstly used for high-dimensional samples when these algorithms are applied in face recognition.Therefore,by using the orthogonal discriminant locality preserving projection algorithm,a direct orthogonal discriminant locality preserving projection algorithm is proposed.Through the corresponding matrix decomposition according to the properties of the Laplacian matrix,the projection matrix can be directly extracted from the original high-dimensional space without using PCA algorithm as the first step.In order to solve the small sample size problem,the null space of the local within-class scatter matrix is obtained and the between-class scatter matrix is maximized.Experimental results on face database demonstrate the effectiveness of the proposed method.