并行多传感器多目标跟踪系统性能优化算法

Centralized Multi-Sensor Structured Branch Multiple Hypothesis Algorithm

研究密集杂波环境下的并行多传感器多目标跟踪优化问题,针对传统的数据互联在杂波数量多,虚警率高的情况下测量与航迹关联计算量大,配对正确率低等缺陷,提出了一种采用多级数据压缩的结构化分枝多假设改进算法,设计了一种确定测量与航迹最优配对的测量聚类方法,将已有的单级数据压缩技术扩展为两级数据压缩。利用上述方法对结构化分支多假设算法中的冗余分枝进行压缩和剪枝,产生航迹家族树最优的"等效分支",并给出优化后的状态估计输出。仿真证明,改进算法减少了运算时间,提高了跟踪精度和实时性,为密集杂波环境下的多目标跟踪提供了一种有效方法。

This paper studies the parallel multi - sensor multi - target tracking in dense clutter. A structured branch multiple hypothesis tracking algorithm based on multi -level data compression is proposed to solve the prob- lem of large association computation, low corrected paring rate between measurements and tracks with traditional data association algorithm performed under the condition of dense clutter and high false alarm rate. The algorithm is de- signed to determine the optimal pairing and expand the existed single - stage data compression to two - stage. Then redundant branches generated in traditional MHT algorithm are processed with this method for compression and prun- ing. After that, Optimal Equivalent branch is remained in track family tree and optimized output state estimate is giv- en. Simulation result shows that this approach performs good tracking accuracy and real -time ability, reduces the computation time and provides an effective way for multi - target tracking in dense clutter.