约束线性描述分析与人脸识别

针对高维,小样本模式识别中的特征提取问题,提出了一种约束线性描述分析方法(CLDA).以线性变换后样本的类内距离与类间距离之比最小作为准则函数,同时加上约束条件使变换后的样本中心沿着特定的正交方向,通过白化变换、Gram-Schimdt正交化和正交子空间投影求解约束准则函数得到最优变换矩阵.针对入脸识别的小样本问题,根据奇异值分解定理实现白化变换.对ORL.和UMIST人脸库进行了仿真研究,结果表明CLDA方法的性能接近于某些Fisher描述分析万法如直接Fisher描述分析(DDA)和改进的Fisher描述分析(R-LDA).

Constrained Linear Discrimination Analysis and Face Recognition

A constrained linear discrimination analysis method was proposed for the feature extraction in the pattern recognition of problems with high dimension and small samples. Applying whitening process and Gram-Schimdt orthogonalization and orthogonal subspace projection, an optimal transformation matrix was designed to minimize the ratio of intra-class distance to inter-class distance while imposing the constraint that different class centers after transformation are along specifically directions that are orthogonal each other. For the small sample problem of face recognition, the whitening process was realized by singular value decomposition. The experimental results using the ORL and the UMIST face image database demonstrate that the effectiveness and performance of CLDA is approximate with some Fisher discrimination analysis such as direct Fisher discrimination analysis (DDA) and regularized Fisher discriminant analysis (R-LDA).