中文摘要：

针对多目标跟踪出现的邻近目标难分辨及计算量大的问题,目前大多数检测跟踪算法在算法性能和计算复杂度之间权衡,不能同时兼顾这两项指标。本文提出了一种基于动态规划的多目标（GMTI-TBD）方法,该算法根据雷达实际检测目标时出现的距离向和方位向展宽,限定目标能量扩展区域,得到新的值函数;在此基础上,通过对目标状态的预测去除多目标的相互干扰。另一方面,利用单次动态规划估计目标所在区域,使轨迹搜索范围大大减小,有效解决了多目标带来的高维计算量问题。仿真结果表明本文算法在检测跟踪性能上优于经典的多目标DP-TBD算法,同时所需计算时间较少。

英文摘要：

In order to distinguish the adjacent targets and reduce the computation amount for tracking multiple targets,most of the current detection and tracking algorithms make a compromise between the performance and computation complexity. However,both of the indices cannot be taken into account simultaneously. Here,a multi-objective GMTI-TBD algorithm based on dynamic programming is proposed for dealing with the multiple targets problem. Extended area of the target energy is restricted through computing range and azimuth extension in the actual radar system,and then the merit function is achieved. Based on this,the mutual interference of adjacent targets can be reduced by predicting the target state. Finally,the searching region is greatly reduced by estimating the target area using single dynamic programming. In this way,the high-dimension calculation problem is effectively solved. Simulation results show that,compared with classic multi-objective DP-TBD algorithms,the proposed method is superior in detecting and tracking the performance,while requiring less computation time.