低秩重检测的多特征时空上下文的视觉跟踪

时空上下文跟踪算法充分的利用空间上下文中包含的结构信息能够有效的对目标进行跟踪，实时性优良。但是该算法仅仅利用单一的灰度信息，使得目标的表观表达缺乏判别性，而且该方法在由于遮挡等问题造成的跟踪漂移后无法进行初始化。针对时空上下文算法存在的弱点，本文提出了一个基于低秩重检测的多特征时空上下文跟踪方法。首先利用多特征对时空上下文进行多方面的提取，构建复合时空上下文信息，充分利用目标周围的特征信息，提高目标表观表达的有效性。其次利用简单有效的矩阵分解方式将跟踪到的历史跟踪信息进行低秩表达，将其引入有效的在线重检测器中来保持跟踪结构的一致稳定性，解决了跟踪方法在跟踪失败后的重定位问题，在一系列跟踪数据集上的实验结果表明本算法比原始算法及当前的主流算法相比有更好的跟踪精度与鲁棒性，且满足实时性要求。

Object Tracking via Low-Rank Redetection Based Multiple Feature Fusion Spatio-Temporal Context Learning

The spatio-temporal tracking (STC) algorithm can effectively track the object using structural information contained in the context around the object in real time. But the algorithm only exploits single gray object feature information so as to make the object representation discriminative. Moreover, it fails to initialize when tracking drift due to occlusion problems. Aiming at the existing weaknesses of the spatio-temporal context algorithm, a novel low-rank redetection base multiple features fusion STC tracking algorithm is proposed in this paper. Firstly, we extract multiple features fusion base spatio-temporal context to construct complicated spatio-temporal context information, and it can improve the effectiveness of object representation by taking full advantage of the feature information around the object. Then we use a simple and effective matrix decomposition method to give a low rank expression of the history of tracking information, which be embedded into the online detector. By that we can keep the uniform structure stability of the tracking algorithm to solve the relocation problem after the tracking failure. Experimental results on a series of tracking Benchmark show the proposed algorithm has a better tracking precision and robustness than several stale-of-the-art methods, and it also have a good real-time requirement.