Object Tracking Algorithm Based on Sparse Representation for Meta-sample
Yin Zhang⋆ · 2014
The research on sparse coding has been introduced into the issue of video object tracking recently. Under particle filter framework,the target template is represented by a set of all target candidates. Trivial templates are used to deal with complex changes of objects in video scenes. In this paper,however,the algorithm ignores the intrinsic information of the target candidates,so the cost is very expensive. This paper proposes an object tracking algorithm based sparse representation for metasample. A set of meta-samples were firstly extracted from all target candidates,then the trivial template were added into building the over complete dictionary. For tracking,an iterative algorithm was proposed to solve l 1-norm minimization. Experimental results indicate that the proposed method is more effective and robust than the existing methods based on l 1 norm minimization.