Modification of Density-Based Clustering and Threshold Adjustment Detection-Tracking Integration Algorithm for 77 GHz Automotive Radar

In this paper, a modification of Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is proposed based on real-world high resolution point clusters, which is obtained by 77-GHz frequency modulated automotive radar through CFAR technology. The modification of DBSCAN includes introducing normalized DBSCAN pre-processing method and a scale factor for dimension stretching/compression, which makes the target have similar measurement in three different dimensions of two-dimensional Cartesian coordinate system and velocity dimension, a The scale factor can adjust the sensitivity of DBSCAN algorithm to velocity dimension to deal with velocity expansion problem. Based on the clustering results, this paper proposes CFAR-JPDA algorithm, which integrates detection and tracking, and proposes a new extended Kalman filter state equation and measurement equation for road moving targets. In terms of data association, the modified JPDA algorithm is proposed, and the shape information of the target is used to confirm the measurement twice, which improves the accuracy of the target association.