基于帧间DBSCAN聚类的毫米波雷达交通多目标跟踪方法

毫米波雷达已成为车联网中的主流传感器之一，可用于交通场景的多目标跟踪。本文将毫米波雷达安装于道路上方进行交通目标跟踪，针对基于帧内DBSCAN(Density-Based Spatial Clustering of Applications with Noise)聚类的多目标跟踪中，在该安装场景下多径噪点难以去除和纵向的交通目标点云难以区分的问题，提出了基于帧间DBSCAN聚类的毫米波雷达交通多目标跟踪方法。该算法使用多帧合并处理的方式，利用帧序特征用于解决多径噪点问题，并利用空间纵向分段的方法改善了原算法在纵向上目标区分度不足的缺点。本文通过六组不同的实际场景实验，证明了本方法在不同场景下，均相比原方法对跟踪结果有不同程度的改善。 

Traffic Multi-object tracking method of millimeter wave radar based on inter-frame DBSCAN clustering

Millimeter wave radar has become one of the mainstream sensors in the Internet of Vehicles, which can be used for multi-target tracking in traffic scenes. In this paper, millimeter-wave radar is installed above the road to track traffic targets. aiming at the problem that multipath noise is difficult to remove and vertical traffic target point clouds are difficult to distinguish in multi-target tracking based on intra-frame DBSCAN (density-based spatial clustering of applications with noise) clustering, a multi-target tracking method of millimeter-wave radar traffic based on inter-frame dbscan clustering is proposed. The algorithm uses multi-frame merging processing, uses frame order features to solve the multipath noise problem, and uses spatial vertical segmentation method to improve the shortcomings of the original algorithm in the vertical target discrimination. In this paper, six groups of experiments in different actual scenes prove that this method improves the tracking results in different degrees compared with the original method in different scenes.