基于聚类与霍夫变换的同型雷达多目标定位算法

传统测向交叉定位手段在定位航迹交错的同型雷达的多目标时，难以正确匹配单个目标的方位参数。针对该问题，提出了一种基于密度聚类与霍夫变换的同型雷达多目标定位及航迹筛选算法。首先，对侦获时间范围进行切片，利用时间切片内的功率与测向参数进行DBSCAN(Density-Based Spatial Clustering of Application with Noise)聚类，依据聚类中心的测向值进行初始定位；而后计算初始定位点到侦察站的距离，利用适用性改进的霍夫变换检测出侦获的功率参数与距离参数变量之间的隐藏线性关系；最后根据提取到的线性参数排除初定位中的虚警点，筛选出单个目标的定位轨迹。仿真结果表明，该算法在面对多达10个轨迹交错的非直线运动目标时仍能有效完成定位计算并筛选出各目标轨迹，保证定位误差和错选率不恶化。

A Multitarget Localization Algorithm for Homogeneous Radar Based on Clustering and Hough Transform

Traditional direction of arrival(DOA) cross location methods are difficult to accurately match the direction parameters of a single target when locating multiple targets of the same type of radar with overlapping tracks. For this problem,a multitarget localization and trajectory screening algorithm based on density clustering and Hough transform for the same type radar is proposed. Firstly, the detection time range is sliced, Density-based Spatial Clustering of Application with Noise(DBSCAN) clustering is performed using the power and direction parameters within the time slice,and initial positioning is carried out based on the direction finding value of the clustering center.Then the distance from the initial positioning point to the reconnaissance station is calculated, and the improved Hough transform is used to detect the hidden linear relationship between the detected power parameters and the distance parameter variables.Finally, according to the extracted linear parameters, the false alarm points in the initial positioning are excluded, and the positioning trajectory of a single target is filtered out. The simulation results show that the algorithm can still effectively complete positioning calculations and screen out the trajectories of up to 10 nonlinear moving targets with overlapping trajectories, ensuring that positioning errors and misselection rates do not deteriorate.