基于卷积神经网络的雷达目标检测方法

雷达目标检测近年来一直是雷达信号处理中的重要任务,在探测监控等安全领域中有非常重要的作用;针对传统恒虚警目标检测方法存在的环境适应能力较弱、复杂地形环境下雷达虚警数量急剧上升等问题,提出一种基于卷积神经网络的雷达目标检测方法;以雷达回波信号数据处理后得到的距离-多普勒图像作为模型的训练集和测试集,设计基于FasterR-CNN结构的雷达目标检测模型,训练模型并将测试结果与传统恒虚警目标检测算法结果相比较,所设计的模型提升了雷达目标检测正确率并较大地减少了虚警数量,这表明将卷积神经网络应用于雷达回波信号的处理任务中是可行的。

Radar Target Detection Based on Convolutional Neural Network

Radar target detection has been an important task in radar signal processing in recent years, and it plays a very important role in detection security fields . Aiming at the problems that traditional constant false alarm target detection methods have weak environmental adaptability and the rapid increase in the number of radar false alarms in complex terrain environments, a target detection method based on convolutional neural networks is proposed. The Range-Doppler image obtained by radar echo data processing is used as the training and test set, and the target detection model is designed based on the Faster R-CNN structure. The training and test results are compared with the results of the traditional constant false alarm target detection algorithm. The model improves the accuracy of target detection and greatly reduces the number of false alarms, which shows that it is feasible to apply convolutional neural networks to radar echo signal processing tasks.