网络协同跟踪下广义分布式航迹关联方法

Generalized distributed track-to-track association algorithm for collaborative network tracking

针对协同网络下的多目标跟踪问题,提出了一种广义分布式航迹关联算法。首先对序贯航迹关联准则进行分析,构造了广义分布式航迹全配对似然函数以及航迹关联统计量;在此基础上,建立广义航迹关联的数学模型,从而将分布式航迹关联转化为多维分配问题;然后利用改进免疫算法来寻求最理想的航迹关联。在航迹关联过程中,先利用chi方分布的假设检验来排除明显不相关的关联组合,再通过计算免疫抗体的适应值来确定多节点航迹间的关联关系;最后利用航迹关联评价指标对所提出的方法进行评估。仿真结果显示,该方法在密集目标环境下具有较好的关联稳定性,与序贯航迹关联方法相比,关联效果得到明显改善。

A generalized distributed track-to-track association method is proposed for multiple targets tracking in collaborative sensor network.Firstly,by analyzing the traditional sequence track-to-track association rules,all pairwise likelihood function and generalized track-to-track association statistic of the distributed tracking are constructed.On the basis of that,the mathematical model of the generalized track-to-track association is established.Then the distributed track-to-track association question is translated into the multiple dimension assignment.Secondly,the optimal association relationship among tracks is searched by using the improved immunity algorithm.The whole track-to-track association course includes two stages.At first,the unrelated track combination is eliminated by using the chi-square distribution hypothesis test,and then the multiple sensors multiple tracks association relationship is confirmed according to the immunity antibody fitness.Finally,the proposed method is evaluated by utilizing some track-to-track association performance parameters.Simulation results show that the association stability and the association effect of the new method are improved obviously compared with the traditional sequence pairwise association method.