A Radar Target Intelligent Detection Algorithm Based on Range Doppler Graph

Radar target detection and recognition is a key aspect in the extensive domain of radar signal processing. The range Doppler map serves as a quantized portrayal of the radar target's motion state and velocity, and it is essential data for the detection and recognition of radar targets. In this paper, an intelligent detection and recognition algorithm for radar targets, grounded on the range Doppler map, is put forward. The algorithm follows a two - stage detection and recognition model. To make the most of the features of the range Doppler map, a sample enhancement strategy involving random transformation and label mixing is utilized. This strategy expands the training samples and boosts the network's robustness. Additionally, the position regression loss is strengthened by integrating IOU and target size, thus enhancing the positioning accuracy. Ablation experiments show that the improvements presented in this paper can effectively enhance the algorithm's detection and recognition performance. Tests on authoritative public datasets reveal that, compared with the existing Faster RCNN and DAROD algorithms, the algorithm proposed in this paper has a remarkable performance improvement.