基于 C3K2‐SC‐YOLOv11的 SAR图像舰船检测
CSTR:
作者:
作者单位:

(1.重庆移通学院公共大数据安全技术重庆市重点实验室,重庆 401420;2.重庆移通学院通信与信息工程学院,重庆 401520;3.重庆医药高等专科学校基础部,重庆 400030)

作者简介:

赖小龙(1987—),男,硕士研究生,副教授,研究方向为图像处理;刘彪(1998—),男,硕士研究生,讲师,研究方向为水下图像处理。

通讯作者:

中图分类号:

TP391

基金项目:

国家自然科学基金项目(62471077);基于深度学习的实时性目标检测研究项目(KY2024011).


Synthetic Aperture Radar Imaging-Based Ship Detection Using C3K2-SC-YOLOv11
Author:
Affiliation:

(1. Chongqing Key Laboratory of Public Big Data Security Technology,Chongqing College of Mobile Communication, Chongqing 401420,CHN;2. College of Communication and Information Engineering,Chongqing College of Mobile Communication,Chongqing 401520,CHN;3. Basic Department,Chongqing Medical and Pharmaceutical College, Chongqing 400030,CHN)

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对传统合成孔径雷达(SAR)舰船图像检测方法精度低、易漏检等,提出一种基于改进 YOLOv11的图像舰船检测方法 C3K2-SC-YOLOv11。首先,设计并行卷积多尺度特征模块 C3K2-SC,以替换 C3K2模块,有效增强了模型对多尺度特征和关键通道的自适应能力。其次,设计 HSAN_PSA模块,以替代 C2PSA,通过深度融合分层空间注意力与位置敏感注意力,有效抑制了相干斑点噪声的干扰。最后,引入 SSSA注意力机制改进 CBAM,在保持 CBAM优势的同时,增强了模型对 SAR图像中小目标的细节捕获能力。在 HRSID数据集上的实验结果表明,相较于较 YOLOv11,改进后的模型平均精度均值 mAP@50和综合平均精度 mAP@50∶95分别提升 6%和

    Abstract:

    This paper presents an imaging-assisted ship detection method based on C3K2-SC-YOLOv11 that overcomes the limitations of low accuracy and easy-to-miss detection in traditional synthetic aperture radar-based ship-image detection methods. First, the parallel convolutional multiscale feature module C3K2-SC was designed to replace the C3K2 module and effectively enhance the adaptive ability of the model. Supported by this new module, the model could simultaneously focus on information-rich channels and important regions to capture salient targets in the global context. Subsequently,the HSAN_PSA module was designed to replace C2PSA. The interference of the coherent speckle noise was significantly suppressed by integrating hierarchical spatial attention with position-sensitive attention. Finally, the SSSA attention mechanism was introduced to improve the CBAM. The SSSA attention maintained the advantages of CBAM while significantly enhancing the performance of the model in small-object detection and multiscale adaptation,making it suitable for object detection tasks in complex scenarios. The experimental results showed that on the HRSID dataset,the proposed model outperformed YOLOv11,showing6% and 9. 59% improvements in mAP@50 and mAP@50∶95,respectively.

    参考文献
    相似文献
    引证文献
引用本文

赖小龙,刘彪,黎希.基于 C3K2‐SC‐YOLOv11的 SAR图像舰船检测[J].半导体光电,2026,47(1):193-202. LAI Xiaolong, LIU Biao, LI Xi. Synthetic Aperture Radar Imaging-Based Ship Detection Using C3K2-SC-YOLOv11[J].,2026,47(1):193-202.

复制
分享
相关视频

文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2025-11-18
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2026-03-20
  • 出版日期:
文章二维码