A New Method of Vehicle Target Radar Data Segmentation Based on DBSCAN and Lane Recognition

Since the raw radar data is always seriously mixed with echoes from multiple targets in a complex traffic environment, it is extremely difficult to segment the target data for feature extraction. In this paper, a new method of vehicle target radar data segmentation based on Density Based Spatial Clustering of Applications with Noise (DBSCAN) and lane recognition is proposed. After signal preprocessing, the vehicle features with lane label are used to train lane recognition model. Then, the coarse segmentation can be obtained by DBSCAN. Finally, the coarse segmentation results are further finely segmented through the above trained lane recognition model. By real-data comparative experiment based on a millimeter-wave (MMW) radar, the effectiveness and superiority of the proposed method for vehicle targets data segmentation in complex traffic environment over the DBSCAN are verified.