针对时空上下文(STC)算法在抗遮挡目标跟踪中的不足，提出使用上下文模型相似度作为判别遮挡的条件和改进上下文模型更新方程的修正系数，同时采用预测算法修正搜索区域，构建了一种基于时空上下文跟踪的抗遮挡目标跟踪算法，并通过标准目标跟踪视频库对原算法和改进后算法的跟踪性能进行仿真和对比。实验证明，在原算法的基础上提高了抗遮挡跟踪的鲁棒性，在一些图像序列中跟踪成功率的提高最高可达30%。

In order to improve the performance of the Spatio-Temporal Context(STC) algorithm in anti-occlusion object tracking, using context model similarity as the judge criterion of occlusion and also the correction coefficient of context model renewal equation, an anti-occlusion object tracking algorithm based on STC tracking is proposed. The simulation and comparison are carried out by the standard target tracking video library. The experimental results show that the robustness of the anti-occlusion tracking is improved on the basis of the original algorithm. In some image sequences, the tracking success rate can be increased by 30%.