一种针对SAR图像的舰船目标检测算法  

A Ship Target Detection Algorithm for SAR Images

作　　者：孟凡龙 齐向阳 范怀涛 MENG Fanlong;QI Xiangyang;FAN Huaitao(Institute of Aerospace Information Innovation,Chinese Academy of Sciences,Beijing 100000,China;School of Electronic,Electrical and Communication Engineering,University of Chinese Academy of Sciences,Beijing 100000,China)

机构地区：[1]中国科学院空天信息创新研究院,北京100000 [2]中国科学院大学电子电气与通信工程学院,北京100000

出　　处：《电光与控制》2025年第1期74-79,共6页Electronics Optics & Control

基　　金：中国科学院基金项目(E2Z216010F)。

摘　　要：由于环境复杂、舰船目标散焦和尺度的多样性,基于SAR图像的舰船目标检测仍然存在一些问题。提出了一种针对SAR图像的舰船目标检测算法。首先,基于可变形卷积构建舰船目标特征细化模块,提高对大长宽比姿态的舰船目标的特征提取能力;其次,在主干网络末尾引入了舰船空间金字塔聚合结构,增强对舰船目标的全局特征提取能力;最后,设计了尺度扩展特征金字塔网络,增强舰船浅层和深层特征信息的交互,提高对多尺度舰船目标的检测能力。实验结果表明,所提算法在HRSID数据集上的mAP达到了93.72%,F1分数达到了89.70%,优于所有比较算法,具有良好的检测效果。Ship target detection based on SAR images continues to face challenges due to environmental complexity,ship target dispersion,and scale diversity.This paper proposes a ship target detection algorithm specifically for SAR images.Firstly,a ship feature refinement module is developed based on deformable convolution to enhance the feature extraction capabilities for ship targets with significant aspect ratios.Secondly,a ship spatial pyramid aggregation structure is integrated at the end of the backbone network,thereby improving the global feature extraction capability for ship targets.Finally,a scale expansion feature pyramid network is designed to facilitate the interaction between shallow and deep feature information of the ship,thereby enhancing the detection capability for multiscale ship targets.Experimental results indicate that the proposed algorithm achieves a mean Average Precision(mAP)of 93.72%and an F1 score of 89.70%on the HRSID dataset,outperforming all the comparative methods and demonstrating effective detection performance.

关 键 词：SAR图像 舰船检测 可变形卷积 舰船空间金字塔聚合结构 尺度扩展特征金字塔网络 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]