检索
E-mail
RSS

作者投稿 主编审稿 专家审稿 远程编辑

热带海洋学报

• • 上一篇    

基于深度学习的SAR图像海洋涡旋自动检测模型

刘太龙1，谢涛1，2，3，4，李建1，3，4，王超1，5，张雪红1，3，4   

  1. 1.南京信息工程大学遥感与测绘工程学院， 江苏 南京 210044；

    2.青岛海洋科技中心区域海洋动力学与数值模拟功能实验室,青岛 山东 266200；

    3.自然资源部遥感导航一体化应用工程技术创新中心,江苏 南京 210044；

    4.江苏省协同精密导航定位与智能应用工程研究中心,江苏 南京 210044；

    5.南京信息工程大学电子与信息工程学院， 江苏 南京 210044）

  • 收稿日期:2024-12-29 修回日期:2025-02-18 接受日期:2025-02-26
  • 通讯作者: 谢涛
  • 基金资助:
    国家重点研发计划资助（2022YFC3104900/2022YFC3104905）、国家自然科学基金（42176180）

Deep Learning-Based Automatic Detection Model for Ocean Eddy in SAR Images

LIU Tailong1, XIE Tao1, 2, 3, 4, LI Jian1, 3, 4, WANG Chao1, 5, ZHANG Xuehong1, 3, 4   

  1. 1.School of Remote Sensing& Geomatics Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, China;

    2. Laboratory for Regional Oceanography and Numerical Modeling, Qingdao Marine Science and Technology Center, 266200, Qingdao, Shandong Province, China;

    3. Innovation Center for Integrated Remote Sensing and Navigation Applications Engineering Technology, Ministry of Natural Resources, Nanjing, Jiangsu 210044, China;

    4. Jiangsu Collaborative Innovation Center for Precision Navigation and Intelligent Application Engineering, Nanjing, Jiangsu 210044, China ；

    5. School of Electronic and Information Engineering, Nanjing University of Information Science and Technology, Nanjing 210044, China）

  • Received:2024-12-29 Revised:2025-02-18 Accepted:2025-02-26
  • Supported by:
    National Key R&D Program of China (2022YFC3104900/2022YFC3104905), National Natural Science Foundation of China (42176180)

     

PDF (PC)

赞

摘要： 传统基于SAR数据的涡旋检测方法需手动设定阈值和特征参数，操作复杂且难以自动化，现有深度学习模型在检测过程中也存在较高的漏检和误检情况，难以满足涡旋检测对精度和效率的要求。为解决这些问题，本文提出一种基于YOLOv8的改进模型，即EddyDetNet，以克服上述局限性，提升检测的准确性和效率。该模型在Backbone和Neck中引入自适应特征融合模块（adaptive feature compression module,AFCM）和多尺度特征空间金字塔模块（multi-scale feature spatial pyramid module, MFSP）并优化Neck结构和在Head部分新增小目标检测头，提升了不同尺度涡旋的检测精度。实验结果表明，EddyDetNet在精确率（precision，P）、召回率（recall，R）和平均精确率均值(mean Average Precision，mAP)指标上较YOLOv8分别提升了2.4%、3.2%和5.5%，参数量和运算量减少了38.1%和15.8%。与YOLOv8相比，EddyDetNet在降低计算复杂度和参数量的同时，保持了较高检测精度，适用于多目标和复杂背景下的涡旋检测任务。

关键词: 海洋涡旋, 合成孔径雷达, 深度学习, YOLO, 目标检测

Abstract: Traditional eddy detection methods based on SAR data require manually setting thresholds and feature parameters, which makes the process complex and difficult to automate. Existing deep learning models also suffer from high false negatives and false positives during detection, failing to meet the accuracy and efficiency requirements of eddy detection. To address these issues, this paper proposes an improved model based on YOLOv8, namely EddyDetNet, to overcome the above limitations and enhance both detection accuracy and efficiency. The model introduces an Adaptive Feature Compression Module (AFCM) in the Backbone and Neck to achieve lightweight and efficient feature extraction, and uses a Multi-scale Feature Spatial Pyramid Module (MFSP) to enhance the effect of multi-scale feature fusion. By optimizing the Neck structure and adding a small-object detection head in the Head part, the model improves the detection accuracy of eddies at different scales. Experimental results show that EddyDetNet outperforms YOLOv8 by 2.4%, 3.2%, and 5.5% in precision (P), recall (R), and mean Average Precision (mAP), respectively, while reducing the parameter size and computational complexity by 38.1% and 15.8%. Compared to YOLOv8, EddyDetNet reduces computational complexity and parameter size while maintaining high detection accuracy, making it suitable for eddy detection tasks in multi-target and complex background scenarios.

Key words: Oceanic eddies, Synthetic Aperture Radar, Deep learning, YOLO, Object Detection