基于稀疏表示的视频目标跟踪研究综述

视频目标跟踪在计算机视觉领域有着广泛应用，由于目标自身和外界环境变化的复杂性和难以预知性，使得复杂场景下鲁棒实时目标跟踪成为一项亟待解决的关键问题.由于视觉信息可以用少量神经元进行稀疏表示，因此稀疏表示已经广泛应用于人脸识别、目标检测和目标跟踪等计算机视觉领域.本文旨在对基于稀疏表示的视频目标跟踪算法进行综述.首先，介绍了基于稀疏表示的视频目标跟踪算法中的字典组成；其次，介绍了稀疏模型的构建及求解算法和模型更新，并对算法复杂度进行了简要分析；然后，对现有公开代码的稀疏表示跟踪算法在测试数据上进行了实验分析，结合算法模型和实验结果对其进行了分析；最后，对基于稀疏表示的视频跟踪算法存在问题进行了讨论，并对未来的研究趋势进行了展望.

Research of Sparse Representation-based Visual Object Tracking: A Survey

Visual object tracking has been widely used in computer vision. Due to the complexity and unpredictability of the object itself and surroundings' changes, robust and real-time tracking is a key issue in urgent need of settlement in complex scenes. Since vision information can be expressed by few neurons, sparse representation has already been used in face recognition, object detection, visual tracking and so on. This paper aims to review the state-of-the-art of sparse representation-based visual tracking algorithms. Firstly, we introduce the codebook employed in the sparse representation-based trackers. Secondly, sparse model construction, corresponding solution and model update are described. And the algorithm complexity is briefly analyzed. Thirdly, the open sparse representation-based trackers' codes are conducted on the benchmark datasets, and the experimental results are fully analyzed in combination with models. Finally, we discuss the existing problems of sparse representation-based trackers and prospect the future.