Light-Weighted Marine Radar Target Detection Based on CFAR-CNN
Aiming at the problems of low resolution and high real-time requirements of coastal defense radar and the difficulty of traditional CFAR algorithm to meet increasingly sophisticated requirements of modern war.In this paper,we incorporate traditional CA-CFAR algorithm into the two-stage target detection framework of computer vision to form a light-weighted and efficient radar target detection algorithm.Firstly,a low threshold CFAR(Lo-CFAR)algo-rithm is applied to indicate whether many positions of potential targets are real targets or false alarms.Then,the data slices are made on the radar echo range-azimuth map according to the position of spots.Finally,a high performance clas-sifier is applied to training data slices.The numerical experiments show that the proposed method has significant advan-tages in improving detection probability,suppressing false alarm and light-weighted timeliness compared with the tradi-tional CFAR and Faster R-CNN algorithm.
国家科技期刊平台
NSTL
万方数据
