基于局部敏感可鉴别组稀疏表示的视频语义分析  

Video semantic analysis based on locality-sensitive discriminant group sparse representation

作　　者：黄小璠 詹永照[1] 柯佳[1] 苟建平[1] 

机构地区：[1]江苏大学计算机科学与通信工程学院,镇江212013

出　　处：《江苏科技大学学报（自然科学版）》2017年第6期814-820,共7页Journal of Jiangsu University of Science and Technology:Natural Science Edition

基　　金：国家自然科学基金资助项目(61170126);江苏省重点研发计划项目(BE2015127);江苏大学高级技术人才科研启动基金项目(13JDG126;14JDG037);江苏省高校自然科学研究面上项目(14KJB520008;14KJB520007)

摘　　要：视频语义分析已经成为计算机领域的研究热点之一.为了提高基于稀疏表示的视频语义分析的准确性,文中结合基于稀疏系数的可鉴别函数和组稀疏模型,分析目前已有的局部敏感稀疏表示字典学习模型,提出一种基于局部敏感可鉴别组稀疏表示.该方法利用组稀疏模型和可鉴别损失函数,在保持数据局部结构的基础上实现样本变量组水平上的稀疏性.所提方法首先构建局部敏感组稀疏字典,然后进行稀疏表示分类.为了验证上述方法的有效性,在TRECVID 2012和OV视频库上,将上述方法与多种算法进行对比,实验结果表明,基于局部敏感的可鉴别组稀疏表示方法能够有效提高视频语义分析的准确率.Video semantic analysis has been a hot research area in computer science. In order to improve the accuracy of video semantic analysis based on sparse representation,we analyze the locality-sensitive sparse representation model,and then integrate discriminant function and group sparse model based on sparse coefficient,which develop a locality sensitive and discriminant group sparse representation（LSDGSR）. In the proposed method,group sparse model and discriminant loss function are introduced into the locality-sensitive sparse representation,which realizes variable group selection ability. We first design an locality-sensitive dictionary,and then do sparse representation classification. In order to verify the effectiveness of the approach,the experiments are conducted on the TRECVID 2012 and OV datasets. The experimental results show that the proposed method can improve the accuracy of video semantic analysis.

关 键 词：视频语义分析 稀疏表示 组稀疏 可鉴别性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]