| 引用本文: | 薛俊杰， 刘良玉， 马小艳， 徐嘉辉. 基于循环随机Hough变换和DBSCAN的群起始算法[J]. 雷达科学与技术, 2025, 23(6): 700-706.[点击复制] |
| XUE Junjie, LIU Liangyu, MA Xiaoyan, XU Jiahui. Group Initiation Algorithm Based on Cyclic Randomized Hough Transform and DBSCAN[J]. Radar Science and Technology, 2025, 23(6): 700-706.[点击复制] |
|
| 摘要: |
| 群目标的航迹起始是群目标跟踪的第一步，常规的航迹起始算法应用在群目标上会产生大量虚假航迹，而传统的群目标起始算法存在抗杂波能力差且未考虑多群重叠的问题。因此提出了一种基于循环Hough变换和基于密度的空间聚类（Density-Based Spatial Clustering of Applications with Noise, DBSCAN）算法的群起始算法。算法通过对多次扫描的点迹做随机Hough变换投影到参数空间，利用群目标运动特性一致的特点通过聚类提取出阈值最大的群，考虑到群的参数积累会影响其他的群或者目标，因此提取完再循环做随机Hough变换依次提取出阈值最大的群直至结束。最后将提取出的群利用DBSCAN算法进行群分割完成群起始。文章最后通过仿真验证，表明该算法不仅有较强的抗杂波能力，同时也能解决密集群的起始难题，且计算量不大，可以在工程上应用。 |
| 关键词: 随机Hough变换 基于密度的空间聚类算法 群起始 密集杂波 |
| DOI：DOI:10.3969/j.issn.1672-2337.2025.06.012 |
| 分类号:TN953；TN957.5 |
| 基金项目:国家部委基金资助项目 |
|
| Group Initiation Algorithm Based on Cyclic Randomized Hough Transform and DBSCAN |
|
XUE Junjie, LIU Liangyu, MA Xiaoyan, XU Jiahui
|
|
Shanghai Aerospace Electronic Technology Institute, Shanghai 201109, China
|
| Abstract: |
| The track initiation of group targets is the first step of group target tracking. The conventional track initiation algorithms will generate a large number of false tracks when applied to group targets. However, the traditional group target initiation algorithm has poor clutter resistance and fail to consider the problem of multi-group overlap. To overcome these limitations, a group initiation algorithm based on cyclic Hough transform and density-based spatial clustering of applications with noise （DBSCAN） is proposed. The algorithm projects the random Hough transform on the multiple scanned points to the parameter space, and uses the characteristics of the consistent motion of the group target to extract clusters with the highest thresholds through clustering. Considering that the parameter accumulation of the group will affect other groups or targets, the random Hough transform is performed to extract the group with the largest threshold until the end. Finally, the extracted groups are segmented by DBSCAN algorithm to complete the group initiation. Simulation results demonstrate that the algorithm not only exhibits strong anti-clutter capability but also effectively resolves the problem of dense groups initiation, with acceptable computational complexity, which can be applied in engineering. |
| Key words: randomized Hough transform DBSCAN algorithm group initiation dense clutter |
