火力与指挥控制2025，Vol.50Issue(4)：93-99,7.DOI:10.3969/j.issn.1002-0640.2025.04.013
基于增强局部特征提取的SAR密集目标检测
SAR Intensive Target Detection Based on Enhanced Local Feature Extraction
摘要
Abstract
Aiming at the low accuracy of SAR image detection in dense areas,an improved algorithm to enhance local feature extraction ability is proposed.The CNeB module is used to enhance the feature extraction of small targets by reducing the spatial dimension and expanding the receptive field of the model.BRA mechanism is introduced to enhance the model's ability to understand the target context information.The loss function WIoU replaces the old CIoU and suppresses the harmful gradients produced by low quality images through intelligent weight adjustment mechanism.Verified on HRSID dataset,mAP increases from 92%to 94.4%,recall rate and accuracy are improved by 4%and 1.1%,respectively,proving that the improved algorithm has significant advantages in intensive target detection.关键词
目标检测/双层路由注意力机制/CNeB模块/合成孔径雷达Key words
target detection/double-layer routing attention mechanism/CNeB module/synthetic aperture radar分类
信息技术与安全科学引用本文复制引用
王杰坤,王金伟,宋富骏,许京新,赵博..基于增强局部特征提取的SAR密集目标检测[J].火力与指挥控制,2025,50(4):93-99,7.基金项目
国家自然科学基金(61201418,62171293) （61201418,62171293）
深圳市基础研究专项基金资助(JCYJ20230808105359045) （JCYJ20230808105359045）
