基于聚类和Hough变换的多编队航迹起始算法

以传感器分辨力较低、编队成员之间距离较近时,不能辨别出编队中的目标数量和队形结构为背景,集中研究了密集杂波下编队的航迹起始问题并分析了该问题的难点所在;在此基础上,首先,通过循环阈值法将前3个时刻的量测点迹分割成多个子群;然后,对分割后的各个子群运用K均值聚类法,更准确地求出各个子群的聚类中心点作为该子群的中心;最后,通过最近邻法,以最大速度为限制将前3个时刻各子群中心点互联,对互联上的中心点经Hough变换,映射到参数空间,从而判别出成功起始的航迹和速度。经仿真验证,该方法对密集杂波环境下编队的航迹起始效果较好,并且速度快,起始率高。

Dense Multi-Formation Track Initiation Algorithm Based on K-Means Clustering and Hough Transform

In this paper, the problem of formation track initiation was studied against a background, which had a dense clutter. Especially when the sensor had a low resolution, which couldn＇t distinguish the number and structure of the formation, then the difficult points were analyzed. In order to initiate the formation tracks, the measurements obtained from the first three frames were separated as several sub-groups using the threshold repetition method at first. Then the k-means clustering method was applied to the sub-groups, the center of the clustering qua the center of. sub-groups. At last, nearest neighbor method was used to associate the center points of sub-groups, the points that were transformed to the parameter space, so the tracks and velocities of the successfully initiated tracks were estimated. The simulation result proved the method had a better effect on dense clutter formation track initiation, and had both short track initiation time and high track initiation probability.