关键词: 稀疏表示, 目标跟踪, 字典学习, 梯度方向直方图

Abstract:  Traditional sparse representation can not effectively distinguish between target and background. Aiming at these problems, a discriminative sparse representation is proposed. It adds a discriminative function to the traditional sparse, thereby greatly reducing the influence of interference factors. An online dictionary learning algorithm based on discrimination sparse representation and [?1] constraint is proposed to upgrade target template. It can effectively reduce the impact of the target and the background of the target template. In addition, Histograms of Oriented Gradient（HOG）feature is used to represent the target. The advantage is its robustness to illumination changes. The proposed tracker is empirically compared with state-of-the-art trackers on some challenging video sequences. Both quantitative and qualitative comparisons show that the proposed tracker is superior and more stable.

Key words: sparse representation, target tracking, dictionary learning, Histograms of Oriented Gradient（HOG）