| 摘要: |
| 算法和数据是影响深度学习技术发展的两大关键因素，大多数学者专注于算法的改进和开拓，仅有少部分学者致力于数据的研究。构建合成孔径雷达（Synthetic Aperture Radar,SAR）图像舰船数据集是SAR舰船目标检测项目的第一步，也是星载SAR图像实际工程应用的基础。分析了影响SAR舰船目标检测性能的关键因素，阐述了SAR舰船数据集的构建方法，概述了TerraSAR-X、“哨兵”1号（Sentinel-1）和高分三号（GF-3）三种SAR图像数据源，并对几种公开的SAR舰船数据集进行梳理与分析,总结了各数据集的发展历程，最后指出构建SAR图像舰船数据集仍需考虑的几个方面。 |
| 关键词: 合成孔径雷达 目标检测 舰船图像 深度学习 数据集构建 数据源 |
| DOI： |
|
| 基金项目: |
|
| Summary of research on construction of SAR image ship target detection dataset |
| HUANG Qiongnan,ZHU Weigang,LI Yonggang |
| (Graduate School,Space Engineering University,Beijing 101416,China;Department of Electronic and Optical Engineering,Space Engineering University,Beijing 101416,China) |
| Abstract: |
| Algorithms and data are the two key factors that affect the development of deep learning technology.Most scholars focus on the improvement and development of algorithms,and only a few scholars are devoted to data research.Constructing the synthetic aperture radar(SAR) image ship dataset is the first step of the SAR ship target detection project,and it is also the basis for the actual engineering application of spaceborne SAR images.This paper analyzes the key factors affecting the performance of SAR ship target detection,expoundes the construction method of SAR ship dataset,summarizes three SAR image data source of TerraSAR-X,Sentinel-1 and GF-3,sorts out and analyzes several public SAR ship dataset,summarizes the development process of each dataset,and finally points out several aspects that still need to be considered when constructing SAR image ship dataset. |
| Key words: synthetic aperture radar(SAR) target detection ship image deep learning dataset construction data source |
