Discriminative Sparse Representation and Online Dictionary Learning for Target Tracking

Traditional sparse representation can not effectively distinguish between target and background. Aiming at these problems, a discriminative sparse representation was proposed, and a discriminative function to the traditional sparse was added for greatly reducing the influence of interference factors. While an online dictionary learning algorithm based on discrimination sparse representation and probabilistic mode was proposed to upgrade target template. It can effectively reduce the impact of the target and the background of the target template. The proposed tracker was empirically compared with state-of-the-art trackers on some challenging video sequences. Both quantitative and qualitative comparisons showed that our proposed tracker was superior and more stable.