基于最大熵模糊聚类的快速多目标跟踪算法研究

A Fast Multi-Target Tracking Algorithm Based on Maximum Entropy Fuzzy Clustering

为了提高杂波环境中多目标跟踪的实时性和精确性,利用最大熵数据模糊聚类方法得到的模糊隶属度表示目标与量测之间的关联概率,同时分析了公共量测对目标的影响,引入影响因子重建互联概率矩阵,结合概率数据关联算法实现多目标的状态估计。该算法避免了对确认矩阵的拆分,解决了联合概率数据关联算法随着目标和回波数目增加而导致的计算量爆炸性增长问题。针对不同杂波密度环境下的临近平行目标和小角度交叉目标的跟踪进行了仿真分析,仿真结果表明:最大熵模糊聚类联合概率数据关联算法是一种有效的快速数据关联算法,在密集杂波环境中跟踪性能依然优于联合概率数据关联算法和经验联合概率数据关联算法,在一定程度上可以避免航迹融合。

In order to improve the real-time and the accuracy of tracking multi-target in dense clutter environment, a new data association algorithm based on maximum entropy fuzzy clustering data association was introduced in this paper, which uses fuzzy membership matrix to express the association probability between target and measurement. At the same time, the effect of the public measurement to the target was analyzed, which uses impact factor to re- construct association probability matrix, combining with the probabilistic data association algorithm to estimate the state of target. This algorithm avoids the splitting the confirmation matrix, so it can solve the problem of the high computational load of the joint probabilistic data association algorithm with the increase of the clutter and target. In addition, simulation and analysis are carried out for tracking parallel targets and cross targets in different clutter density environment. Simulation shows that this algorithm is an efficient and fast data association algorithm, and the tracking performance is superior to the joint probabilistic data association algorithm and the cheap joint probabilistic data association algorithm in dense clutter environment, and also can avoiding the tracking coalescence to a certain extent.