基于高斯混合EM聚类的多编队航迹起始方法

针对现有航迹起始方法难以对编队目标进行有效航迹起始的问题,在Hough变换法及其衍生算法基础上,提出基于Hough变换和高斯混合最大期望(expactation maximazation,EM)聚类的多编队目标航迹起始方法。该方法首先利用量测数据的时序信息和目标的运动参数进行筛选,剔除大量虚假量测;再对筛选后的量测数据进行Hough变换,得到初步航迹信息;然后利用相异度矩阵对所得航迹进行预聚类,完成聚类中心初始化;最后进行高斯混合EM聚类,得到聚类结果。仿真结果表明,与Hough变换法及其衍生算法相比,该方法能够快速有效地起始编队目标的航迹,解决了目标密集带来的航迹起始混乱问题。

Multi-formation track initiation method based on Gaussian mixture EM clustering

Focusing on the difficulty in effective track initiation of formation targets with existing track initiation methods, a method of multi-formation target track initiation based on Hough transformation and Gaussian mixture expactation maximazation (EM) clustering is proposed on the basis of Hough transformation and its derivation algorithm. Firstly, the timing information of the measured data and the motion parameters of the target are used to screen out a large number of false measurements. Then the preliminary track information can be obtained by Hough transformation on the measured data after screening. Then the obtained track is pre-clustered by using the phase difference matrix, and the clustering center initialization is completed. Finally, Gaussian mixture EM clustering is carried out to obtain the clustering results. The simulation results show that compared with Hough transformation and its derivation algorithm, this method can quickly and effectively start the track of formation targets, and solve the track starting confusion problem caused by dense targets.