面向合成孔径雷达图像任意方向舰船检测的YOLOv3模型的改进

Improvement of YOLOv3 model for ship detection in any direction in synthetic aperture radar image

为了对舰船图像进行有效检测并及时输出合成孔径雷达(Synthetic Aperture Radar,SAR)图像的目标位置和方位角信息,文章以改进YOLOv3的任意方向为基础,构建舰船目标检测模型。文章主要分析了面向合成孔径雷达图像技术的应用现状,探索了YOLOv3模型算法的损失函数以及数据集构建情况,以掌握YOLOv3模型的改进思路,模拟分析YOLOv3模型的输入范围。经过研究可以确定,改进模型可应用于纯海洋背景下的SAR图像目标检测过程,可以满足监察目标检测的实时性需求。

In order to effectively detect the ship image and output the target position and azimuth information of synthetic aperture radar(SAR)image in time,this paper constructs the ship target detection model based on improving the arbitrary direction of YOLOv3.This paper mainly analyzes the application status of SAR image technology,and explores the loss function of YOLOv3 model algorithm and the construction of data set,so as to master the improvement idea of YOLOv3 model and simulate and analyze the input range of YOLOv3 model.After research,it can be determined that the improved model is suitable for the target detection process of SAR image in pure ocean background,and can meet the real-time requirements of monitoring target detection.