基于流形学习的整体正交稀疏保留鉴别分析

Analysis of Preserving Discriminant of Holistic Orthogonal Sparsity Based on Manifold Learning

稀疏保留投影是一种有效的特征提取方法,但是其主要关注样本间的全局稀疏重构关系,并且得到的投影变换通常不是正交的。在实际应用中,图像数据往往处于高维空间中的一种低维流形中,正交性一直被认为有利于提高鉴别能力。文中以有监督学习的方式在稀疏保留投影中引入了流形结构保留,并使得投影空间正交,从而提出了一种新的特征提取方法,即基于流形学习的整体正交稀疏保留鉴别分析(MLHOSDA)。在人脸和掌纹图像数据库的实验结果表明此方法具有较好的识别效果。

Sparsity Preserving Projections（ SPP） is an effective feature extraction method.However,it focuses on the global sparse reconstruction relations among samples,and its achieved transformation is usually not orthogonal.In real application,image samples possibly reside on a nonlinear submanifold of the high-dimensional space,which is the inherent structure among the samples,and orthogonality is favorable for classification in many scenarios.In this paper,propose a new feature extraction approach named Manifold Learning based Holistic Orthogonal Sparsity preserving Discriminant Analysis（ MLHOSDA）,which introduces the manifold preserving into SPP in a supervised learning manner and makes the obtained transformation orthogonal.The experiment results on face and palmprint image databases demonstrate the effectiveness of the proposed approach.