基于多线索融合的鲁棒粒子滤波跟踪算法

A ROBUST TRACKING ALGORITHM FOR PARTICLE FILTERING BASED ON MULTI-CUE INTEGRATION

为了提高目标特征的表达能力和跟踪的鲁棒性,提出基于多线索的目标跟踪算法。该算法分别从目标颜色特征和结构特征来考虑,在此基础上提出了融合公式,从而使目标在不同场景中都能自适应变化,以提高跟踪的精度和性能,最后,通过几组仿真实验对该算法进行了验证。实验结果表明,该算法对于部分遮挡等情况具有良好的鲁棒性和跟踪精度。

In order to improve the robustness of object tracking and the capability to express the characteristics of the object,this paper presents a robust object tracking algorithm based on multi-cue integration. The integration formula is contrived based on the object colour mode and structural features respectively, thus the objectives can change adaptively in different scenarios, so that the tracking accuracy and performance are well improved. The algorithm has been verified in computer by simulation experiments and the results indicate its good robustness and tracking accuracy in condition of partial occlusion.