Robust Object Tracking via Multi-block and Sparsity-Based Representation

Abstract

Abstract-In this paper, we propose a robust object tracking algorithm based on classifier, multi-block and local sparse coding with multiple discriminative dictionary. The first part of the propose method is train a classifier with the dictionary encodes the information of both target information and background information. The second part exploits the block information of the object target. The different blocks in a sample should contribute differently to the visual tracking, the model effectively exploit the similarity and distinctiveness of different blocks. Each block is coded on its own discriminative dictionary to allow flexible coding block and the parameter after sparse coding can be used for the weights allocation simultaneously. Furthermore, the update scheme considers both the latest observations and the original template, thereby enabling to alleviate drift problem. Extensive experiments on challenging sequences show that the robust tracking achieved by our algorithm.Index Terms-Visual tracking, sparse representation, multi-block.