A Density-Based Adaptive Distance Fuzzy Clustering Algorithm Based on the Multi-target Traffic Radar

Abstract

In the multi-target traffic radar scene, the clustering accuracy between vehicles with close driving distance is relatively low. To solve this problem, this paper proposes a new clustering algorithm, namely a density-based adaptive distance fuzzy (DB-ADF) clustering algorithm. Firstly, the adaptive distance is used to calculate the similarity between points. Secondly, the neighborhood radius clusters are adaptively searched and discovered, which is the result of the initial clustering through cluster merging. Finally, the algorithm takes the result of the first clustering as the input of the second clustering, and iterates over the membership matrix and cluster center to obtain the clustering result. In the experiment, the proposed algorithm was run on the real radar datasets. The clustering performance of the DB-ADF algorithm was compared with the fuzzy c-means (FCM), Gustafson-Kessel (GK), and the density-based spatial clustering of applications with noise (DBSCAN) algorithm. The results show that the DB-ADF algorithm has higher clustering accuracy for the real radar data in some short-range vehicle scenarios.