针对目标跟踪中如何有效地描述目标, 以适应跟踪环境变化的问题, 提出了一种基于分布域描述算子的视频目标跟踪算法。首先, 采用分布域描述算子来建立目标模型; 然后选择粒子滤波算法进行目标状态估计实现目标跟踪; 在跟踪过程中, 采用在线更新模型的方法, 降低了光照变化以及部分遮挡带来的影响, 从而提高跟踪的准确率。在常用的视频序列上进行了实验, 并与目前常用的算法进行了比较。实验结果表明该算法具有较好的跟踪效果, 能实现复杂场景下的目标跟踪。

A novel visual object tracking method based on distribution fields descriptor is proposed to solve the problem how to describe the object effectively and adapt to the change of environmental in visual tracking. Firstly, a distribution fields descriptor is used to build the object model. Then, the object is tracked by particle filter algorithm. The model is updated online. It can reduce the impact of varying illumination and partial occlusion, thereby enhancing the efficiency of object tracking. The proposed algorithm runs in real time and performs favorably against state-of-the-art algorithms on challenging sequences. The experimental results indicate that the proposed algorithm is effective for tracking and has good performance in complex scene.