鲁棒低秩稀疏表示的在线目标跟踪  

Online object tracking with robust low-rank sparse representation

作　　者：陈芸[1,2] 吴飞[2] 荆晓远[2] 

机构地区：[1]江苏信息职业技术学院物联网工程系,江苏无锡214153 [2]南京邮电大学自动化学院,江苏南京210003

出　　处：《计算机工程与设计》2016年第4期1062-1066,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(61272273;61170305);江苏省普通高校研究生科研创新计划基金项目(CXLX13_465);江苏省高等职业院校国内高级访问学者计划基金项目(2014FX034)

摘　　要：为提高目标跟踪的准确性,针对当前目标跟踪算法因光照、遮挡以及姿态变化等因素引起的漂移问题,提出一种鲁棒低秩稀疏表示的在线目标跟踪算法(LRSP)。以粒子滤波作为目标跟踪的基本框架,通过联合采用低秩矩阵恢复和稀疏表示,发现连续帧和密集粒子之间潜在结构信息,降低数据维度,减少计算复杂度,提高目标跟踪的准确性。实验结果表明,相对于其它目标跟踪算法,LRSP算法可以更准确地跟踪目标,对光照和姿态变化具有良好的鲁棒性,对于严重遮挡目标跟踪问题具有明显优势。To improve the accuracy of target tracking,for solving drift problems caused by illumination,occlusion,pose variation and other factors,an algorithm of low-rank robust sparse representation of the target online tracking（LRSP）was proposed.Using particle filter algorithm as a basic framework and jointing low rank matrix recovery and sparse representation,the underlying relationship between consecutive frames and dense particles was found.The dimension of data and the computational complexity were reduced,and the accuracy of target tracking was improved.Experimental results show that,compared with other target tracking algorithms,the LRSP algorithm can track the target more accurately,and has good robustness to illumination and pose variation.The algorithm has obvious advantages in more serious occlusion target tracking cases.

关 键 词：低秩 稀疏 粒子滤波 目标跟踪 鲁棒主成分分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]