中文摘要：

针对多机动椭圆目标的跟踪问题进行研究，建立一种非线性跳变马尔科夫系统模型，继而提出一种新的扩展目标高斯混合概率假设（ET-GM-PHD）滤波算法。新算法引用无损变换（UT）法，解决模型的非线性性，采用一种基于距离的启发式分区策略对观测集合进行了分区，该分区策略简便易操作而且能很大程度地降低算法的计算量。仿真实例表明：新算法比未进行分区的标准GM-PHD滤波算法更加强健有效。

英文摘要：

The problem of tracking multiple maneuvering elliptical targets was studied and a nonlinear jump Markov system model was established. Then a new Gaussian mixture PHD filter for extended targets （ET-GM-PHD） was proposed. This ET-GM-PHD filter adopted the unscented transform technique to overcome the nonlinearity of model, and was carryed out by using a heuristic measurement partition scheme based on the distances between the measurements. This partition scheme is easy, operable and can greatly reduce the computation of the algorithm. Simulation results show that the proposed algorithm is more effective and robust than the standard GM-PHD filter without measurement partition.