数据关联是多目标跟踪的一项关键技术。JPDA是大家公认的多目标跟踪中性能较好的数据关联算法，它认为量测和目标是一一对应的关联关系，但在许多实际情况中，量测和目标是多-多对应的关系。针对上述情况，该文提出了广义概率数据关联算法（Generalized Probability Data Association，GPDA）。文中从理论上对这两种算法的性能进行了详细分析，并利用Monte Carlo技术对其性能进行了仿真比较。

Data association is one of the key technologies in multi-target tracking. And JPDA is considered as the best data association method. JPDA considers the association of measurements with targets is simply one-to-one problem. But in many practical cases, the association of measurements with targets will be multiple-to-multiple problem. For this case, a Generalized Probability Data Association (GPDA) algorithm is proposed in this paper. Furthermore, this paper analyzes the performance of these two algorithms theoretically. And we give the comparative analysis of those performances by using Monte Carlo method.

TP273

国家自然科学基金资助项目（60172037）.