稀疏保持典型相关分析及在特征融合中的应用  

Sparsity Preserving Canonical Correlation Analysis with Application in Feature Fusion

作　　者：侯书东[1] 孙权森[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,南京210094

出　　处：《自动化学报》2012年第4期659-665,共7页Acta Automatica Sinica

基　　金：国家自然科学基金(60773172);教育部博士学科点基金(200802880017);江苏省自然科学基金(BK2008411)资助~~

摘　　要：稀疏保持投影(Sparsity preserving projections,SPP)由于保持了数据间的稀疏重构性,因而获取的投影向量满足旋转、尺度和平移的不变性,并能够在无标签的情况下提取样本的自然鉴别信息,在人脸识别领域取得了较为成功的应用.本文在典型相关分析(Canonical correlation analysis,CCA)的基础上引入稀疏保持项,提出一种稀疏保持典型相关分析(Sparsity preserving canonical correlation analysis,SPCCA).该方法不仅实现了两组特征集鉴别信息的有效融合,同时对提取特征间的稀疏重构性加以约束,增强了特征的表示和鉴别能力.在多特征手写体字符集与人脸数据集上的实验结果表明,SPCCA比CCA具有更优的识别性能.Sparsity preserving projections（SPP） aim to preserve the sparse reconstructive relationship among the data and have been successfully applied in face recognition.The projections are invariant to rotations,rescalings,and translations of the data,and more importantly,they contain natural discriminating information even without class labels.Enlightened by this,we propose a sparsity preserving canonical correlation analysis（SPCCA）.It can not only fuse the discriminative information of two feature sets efficiently,but also constrains the sparse reconstructive relationship among each feature set in order to increase the representational power and has good discrimination capability of the feature extracted by SPCCA.Experimental results on the multiple feature databases and face databases show that the proposed SPCCA is better than CCA.

关 键 词：典型相关分析(CCA) 稀疏保持投影(SPP) 稀疏保持典型相关分析(SPCCA) 特征融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]