一种改进的汽车雷达数据实时聚类算法

An Improved Real-time Clustering Algorithm for Automotive Radar Data

为了更好地在复杂多目标环境下进行汽车雷达数据的实时聚类,使用扩展卡尔曼滤波算法(EKF)对基于密度的聚类算法(DBSCAN)进行改进,并通过仿真和实测实验进行验证。结果表明:新算法在进行增量聚类时每次耗时可以保持在一个稳定且较低的水平;新聚类在不增加时间复杂度的情况下进行自适应聚类,可以解决汽车雷达数据密度不均匀的情况。可见新算法同时实现了增量和自适应DBSCAN聚类,同时保证聚类的效率和准确度。

In order to better perform real-time clustering of automotive radar data in a complex multi-target environment,the extended kalman filtering algorithm(EKF)is used to improve the density-based spatial clustering of applications with noise(DBSCAN)and verified by simulation and actual experiments.The results show that the new algorithm can maintain a stable and low level each time in incremental clustering;The new clustering performs adaptive clustering without increasing the time complexity,which can solve the problem of uneven data density of automobile radar.It can be seen that the new algorithm realizes both incremental and adaptive DBSCAN clustering,while ensuring the efficiency and accuracy of clustering.