基于稀疏表示的目标跟踪方法  

Target Tracking based on Sparse Coding

作　　者：张盛平[1] 姚鸿勋[1] 孙鑫[1] 卢修生[1] 

机构地区：[1]哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《智能计算机与应用》2013年第1期21-25,共5页Intelligent Computer and Applications

基　　金：国家自然基金(61071180)

摘　　要：提出了一个新的基于稀疏表示的目标跟踪方法。在粒子滤波框架下,将目标模板线性表示为所有目标候选的线性组合。当假设目标候选中存在与目标模板相似的候选时,线性表示的系数满足稀疏性约束,可以通过L1范式最小化求解。每一个目标候选在线性表示中的系数反映了该候选与目标模板的相似程度,因此可以将系数作为目标候选的权重。目标跟踪的结果为权重最大的候选。实验结果表明本文提出的算法比文献中现有的基于L1范式最小化的跟踪方法性能更稳定、计算效率更高。This paper proposes a novel target tracking method based on sparse coding. Under particle filter framework, the target template is represented by a set of all target candidates. Assuming that there is one target candidate similar to the target template, the representa- tion coefficients are sparse ,therefore can be solved by L1 norm minimization. The coefficient associated to each target candidate reflects the similarity between the template and the candidate. Therefore, the coefficient can be used as weight of the target candidate. The tracking result is the candidate with the largest weight. Experimental results indicate that the proposed method is more effective and efficient than the existing methods based on L1 norm minimization.

关 键 词：目标跟踪 稀疏表示 L1范式最小化 

分 类 号：TP391.2[自动化与计算机技术—计算机应用技术]