SAR图像智能解译样本数据集构建进展综述

A review on the construction of sample datasets for intelligent interpretation of SAR images

随着深度学习技术在合成孔径雷达(Synthetic Aperture Radar,SAR)图像智能解译领域的不断应用与发展,其对样本数据集的规模、质量、泛化能力等提出了更高要求。近年来,国内外众多研究机构相继发布了一系列SAR样本数据集,但是目前尚缺乏对上述数据集的综合分析。面向SAR图像智能解译需求,首先,梳理了现有的SAR样本采集流程。其次,归纳总结了已公开的SAR样本数据集,并针对现有数据集样本数量有限,探讨了数据扩充方法。最后,讨论了目前SAR数据集构建仍然存在的问题及未来展望。文章首次对SAR样本数据集进行了综述研究,可为SAR样本库建设提供参考。

With the continuous application and development of deep learning technology in the field of intelligent interpretation of Synthetic Aperture Radar(SAR)images,it puts forward higher requirements for the scale,quality and generalization ability of sample data sets.In recent years,many research institutions at home and abroad have released a series of SAR sample datasets,but there is still a lack of comprehensive analysis of the above datasets.In order to meet the requirements of intelligent interpretation of SAR images,firstly,the existing SAR sample collection process is reviewed in this paper.Secondly,the published SAR sample datasets are summarized,and the data augmentation methods are discussed in view of the limited number of samples in the existing datasets.Finally,the existing problems and future prospects of SAR dataset construction are discussed.This paper provides a review study on the SAR sample dataset for the first time,which can provide a reference for the construction of the SAR sample database.