文中提出了一种利用公路网先验知识提高GMTI跟踪效果的新算法。公路网中的目标跟踪问题实质上是一个向量空间受约束的非线性估计问题。利用公路网这个约束条件，建立了一种新的目标状态模型，并利用适合于非线性非高斯状态的UPF算法进行跟踪滤波。仿真实验表明该算法降低了粒子滤波的计算量，提高了目标跟踪的精度。

A new approach to improve the performance of GMTI tracking using prior knowledge of road networks is presented. The problem of tracking an on-road target and a constrained nonlinear estimation problem are consistent essentially. Based on road networks constrained information, dynamic model is built, and a nonlinear/non-gaussian filter technique-UPF is introduced for filtering. Simulation results show that the proposal algorithm can improve the performance of tracking while reducing the computational requirements.

TP953