测绘学报 ›› 2026, Vol. 55 ›› Issue (10): 1814-1828.doi: 10.11947/j.AGCS.2026.20260182

• 海洋测绘 • 上一篇    下一篇

双曲线形态先验引导的浅剖海底管道探测

张胤博1(), 李邵波1,2,3,4(), 张毅4,5, 吴云龙1,4   

  1. 1.中国地质大学(武汉)地理与信息工程学院,湖北 武汉 430074
    2.浙江大学杭州国际科创中心海洋精准感知技术全国重点实验室,浙江 杭州 311215
    3.自然资源部第二海洋研究所海底科学与划界全国重点实验室,浙江 杭州 310012
    4.中国地质大学(武汉)信息技术湖北省重点学科实验室,湖北 武汉 430074
    5.中国地质大学(武汉)地质探测与评估教育部重点实验室,湖北 武汉 430074
  • 收稿日期:2026-05-12 修回日期:2026-10-19 出版日期:2026-11-16 发布日期:2026-11-16
  • 通讯作者: 李邵波 E-mail:1202521367@cug.edu.cn;lishaobo@cug.edu.cn
  • 作者简介:张胤博(2002—),男,硕士生,研究方向为海底声呐目标探测。E-mail:1202521367@cug.edu.cn
  • 基金资助:
    国家自然科学基金(42304049; 42574073);中国博士后科学基金(2023M743282; 2024T170856);湖北省自然科学基金(JCZRMS202600020);海底科学与划界全国重点实验室资助项目(KLSG2406);海洋精准感知技术全国重点实验室资助项目(OSKF-2025M07)

Submarine pipeline detection using sub-bottom profiler guided by hyperbolic morphological priors

Yinbo Zhang1(), Shaobo Li1,2,3,4(), Yi Zhang4,5, Yunlong Wu1,4   

  1. 1.School of Geography and Information Engineering, China University of Geosciences, Wuhan 430074, China
    2.State Key Laboratory of Ocean Sensing, ZJU-Hangzhou Global Scientific and Technological Innovation Center, Zhejiang University, Hangzhou 311215, China
    3.State Key Laboratory of Submarine Geoscience, Second Institute of Oceanography, Ministry of Natural Resources, Hangzhou 310012, China
    4.Hubei Key Laboratory of Information Technology, China University of Geosciences (Wuhan), Wuhan 430074, China
    5.Key Laboratory of Geological Survey and Evaluation, Ministry of Education, China University of Geosciences (Wuhan), Wuhan 430074, China
  • Received:2026-05-12 Revised:2026-10-19 Online:2026-11-16 Published:2026-11-16
  • Contact: Shaobo Li E-mail:1202521367@cug.edu.cn;lishaobo@cug.edu.cn
  • About author:Zhang Yinbo (2002—), male, postgraduate, majors in seafloor sonar target detection. E-mail: 1202521367@cug.edu.cn
  • Supported by:
    The National Natural Science Foundation of China(42304049; 42574073);China Postdoctoral Science Foundation(2023M743282; 2024T170856);The Natural Science Foundation of Hubei Province(JCZRMS202600020);The Open Project of the State Key Laboratory of Submarine Geoscience(KLSG2406);The Open Project of the State Key Laboratory of Ocean Sensing(OSKF-2025M07)

摘要:

海底管道是海洋油气等物质传输的重要载体,基于浅地层剖面声呐(浅剖)进行管道探测以获取掩埋位置、状态等信息是海底管道维护的重要步骤。现有浅剖管道探测方法对形态先验特征利用不足,在高噪声、地层背景复杂的环境下难以实现精确可靠的探测。针对这一问题,本文提出一种嵌入双曲线形态特征的浅剖管道探测模型。首先,通过构建水平、垂直与斜向协同处理的曲线型卷积模块,捕捉管道目标的曲线状结构特征,并集成多尺度特征聚合模块实现尺度自适应,以抑制特征提取过程中的散射噪声干扰。然后,分析目标双曲线形态的拐点等关键点,建立关键点引导的精细目标检测头,结合传统目标边界框的粗定位,实现目标的粗-精协同探测。最后,引入Inner-IoU损失函数,以提升复杂环境下的识别精度。实测数据测试结果表明,本文模型取得了0.813 7的召回率和0.823 8的F1值,以及0.826 0的mAP值,均优于现有对比方法,可为海底管道自动化探测提供有效方案。

关键词: 海底管道探测, 浅地层剖面仪, 动态蛇形卷积, YOLOv11, 形态特征

Abstract:

Submarine pipelines are a critical component of offshore oil and gas and communication networks, and detection using sub-bottom profiler (SBP) is an essential method for determining the burial status of pipelines. Existing SBP image interpretation methods do not fully utilize the a priori morphological features of pipelines, making it difficult to achieve accurate and reliable detection in high-noise environments with complex geological backgrounds. To address this issue, this paper proposes a SBP pipeline detection model that incorporates hyperbolic morphological features. First, by constructing a curved convolutional module that performs horizontal, vertical, and oblique collaborative processing, the model captures the curved structural features of pipeline targets. It also integrates a multi-scale feature aggregation (MSFA) module to achieve scale adaptability, thereby suppressing scattering noise interference during feature extraction. Second, by analyzing key points such as inflection points in the target's hyperbolic shape, the method establishes a keypoint-guided fine-grained target detection head. Combined with the coarse localization provided by traditional target bounding boxes, this approach achieves coordinated coarse-to-fine target detection. Finally, an Inner-IoU loss function is introduced to improve recognition accuracy in complex environments. Test results on real-world data show that the proposed model achieved a recall of 0.813 7, an F1 score of 0.823 8, and an mAP of 0.826 0, all of which outperform existing comparison methods, providing an effective solution for the automated detection of submarine pipelines.

Key words: submarine pipeline detection, sub-bottom profiler, dynamic snake convolution, YOLOv11, morphological characteristics

中图分类号: