针对复杂环境下多运动目标跟踪的高冲突信息组合问题, 在 DSmT的理论框架下, 提出了一种将局部冲突在局部进行分配的改进证据合并方法来研究多运动目标的跟踪问题。该方法通过深入分析冲突及其相关因素之间的关系, 将目标的位置信息与颜色信息进行融合建立起多信息合并的运动目标跟踪模型。仿真实验结果表明: 采用该方法可实现互遮挡条件下的多目标跟踪, 大大提高了复杂环境下多运动目标跟踪的有效性和稳定性, 对于高冲突证据能够获得良好的跟踪效果。

In order to analyze the combination of high conflict information of moving target tracking in complicated environment, a modified combination rule of evidence by distributing local conflict to local focal element is proposed in the framework of Dezert Smarandache Theory (DSmT). Based on thorough theoretic analysis on the relations of conflict and correlative factors, location and color cues of moving targets were combined by the introduced method, and the tracking model of combined multi-source information was established. Experimental results show that the target tracking is realized under occlusion conditions by the method, the efficiency and reliability of multiple video moving target tracking are improved greatly, and the robust tracking results can be obtained, which has high evidence conflict.