Robust object tracking based on discriminative analysis and local sparse representation

To improve robustness in cases of partial occlusion, deformation and rotation in visual tracking, local similarity measurements are usually used. However, this method have drawbacks under complex backgrounds. For example, the method only consider the traditional similarity measurements of objects and templates, results in the matching errors are prone to lead to the failure of tracking. In this paper, we proposes a object tracking algorithm based on measurements of the local discriminative similarities. This new method have advantages as following: firstly, both the similarities and the discrimination are considered; Secondly, the discriminative weight learning of the local region is carried out to improve the accuracy of fragment measurement; At last, an effective and efficient tracker is designed based on the difference analysis and a simple update manner within the particle filter framework. Experimental results show that the proposed algorithm achieves better performance than traditional competing methods.