基于群目标的多目标关联算法研究

Study of Association Algorithm Based on Group-target for Multi-targets

针对群目标跟踪数据关联的特点以及多目标跟踪数据关联经典的联合概率数据关联算法存在的计算量大与假设条件苛刻等问题,提出了基于群目标的多目标概率数据关联算法GJPDA(Group-target joint probability data association)。该算法首先把跟踪空间内的所有回波看作一个群,跟踪空间中任一关联门内的所有回波看作一个子群,通过关联门是否交叉、多少回波位于关联门交叉区内的判别及其相对于不同关联中心概率的计算,确定交叉区域内回波的归属。以每一个关联门内所有回波的等效回波为量测实现多群目标跟踪。仿真结果表明了该算法的有效性。

Group-target tracking association is a special problem comparing with ordinary mupti-target track association. Meanwhile, when data association is achieved in JPDA algorithm, the computational amount will increases in index following target＇s amount increasing, and its presupposition is rigorous. Then Association Algorithm Based on Group- target for Multi-targets tracking （GPDA） was proposed. Firstly, all echoes in the tracking space were regarded as a group; secondly, all echoes in the association gate were treated as a subgroup; thirdly, whether or not association gates being across was judged by both of measurements and predict value, and how many echoes were located at the across area and how much every echo＇s probability was for its center of the association gate. The simulation results validate its feasibility.