基于有效量测分集的联合概率数据互联算法

Joint Probabilistic Data Association Based on Subsets of Validated Measurements

杂波环境下用联合概率数据互联算法(JPDA)跟踪多目标,其计算量将随跟踪目标数的增多和杂波密度的增大而呈指数增长,因此实时性不强;并且JPDA跟踪杂波环境中的近距离目标时,容易造成航迹合并。在充分考虑单个目标独立性的基础上,提出有效量测分集概念。将接收到的有效量测信号按照单个目标的关联划分,确定各目标跟踪波门内的候选量测信号,不考虑量测信号的重复关联。取得单个目标独立性之后,再运用单目标概率数据互联的方法估计目标状态。仿真实验表明,较传统JPDA实时性更强,能够分辨并跟踪近距离目标。

When using Joint Probabilistic Data Association（JPDA） for multi-target tracking,the calculating amount would increase exponentially with the increasing of number of targets and the density of clutter,thus its real-time performance is not satisfactory.Besides,as tracking closely-spaced targets in the clutter scenario,JPDA could easily lead to track coalescence.A different approach based on the complete independence of targets was presented.The approach regarded all the valid measurements gained from one scan as a whole set,divided it into numerical subsets responding to targets one-by-one.Each subsets contained all the valid measurements associated with their responded targets,regardless been contained by the other one,and estimated the states of targets based on the Probabilistic Data association filter algorithm.Simulations verified that：compared with traditional JPDA,the presented algorithm has better real-time performance and can resolve the closely-spaced targets.