测绘通报 ›› 2023, Vol. 0 ›› Issue (8): 34-39,90.doi: 10.13474/j.cnki.11-2246.2023.0228

• 学术研究 • 上一篇    下一篇

基于激光点云灰度图像的隧道渗水病害检测

周宝定1,2,3, 谢沛瑶1,2,3, 郭文浩4,5, 毛庆洲6   

  1. 1. 深圳大学土木与交通工程学院, 广东 深圳 518060;
    2. 深圳大学城市智慧交通与安全运维研究院, 广东 深圳 518060;
    3. 滨海城市韧性基础设施教育部重点实验室, 广东 深圳 518060;
    4. 西南交通大学地球科学与环境工程学院, 四川 成都 611756;
    5. 深圳大学广东省城市空间信息工程重点实验室, 广东 深圳 518060;
    6. 武汉大学遥感信息工程学院, 湖北 武汉 430079
  • 收稿日期:2022-11-08 发布日期:2023-09-01
  • 通讯作者: 毛庆洲。E-mail:qzhmao@whu.edu.cn
  • 作者简介:周宝定(1986-),男,博士,副教授,主要研究方向为室内定位。E-mail:bdzhou@szu.edu.cn
  • 基金资助:
    广东省自然科学基金(2021A1515011468)

Detection of tunnel leakage based on gray scale image of laser point cloud

ZHOU Baoding1,2,3, XIE Peiyao1,2,3, GUO Wenhao4,5, MAO Qingzhou6   

  1. 1. College of Civil and Transportation Engineering, Shenzhen University, Shenzhen 518060, China;
    2. Institute of Urban Smart Transportation & Safety Maintenance, Shenzhen University, Shenzhen 518060, China;
    3. Key Laboratory for Resilient Infrastructures of Coastal Cities(Shenzhen University), Ministry of Education, Shenzhen 518060, China;
    4. Faculty of Geosciences and Environmental Engineering, Southwest Jiaotong University, Chengdu 611756, China;
    5. Guangdong Key Laboratory of Urban Informatics, Shenzhen University, Shenzhen 518060, China;
    6. School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079, China
  • Received:2022-11-08 Published:2023-09-01

摘要: 地铁盾构隧道渗水检测是隧道维护的基础。针对传统的检测手段难以在隧道内部光线昏暗条件下获得高质量病害图片的问题,本文使用了激光雷达对隧道进行内部渗水检测。首先,基于激光点云灰度图中的渗水病害特征,建立了激光点云灰度图像渗水病害数据集;然后,以Mask R-CNN(Region-CNN)模型为基准框架,采用Swin transformer网络作为底层特征提取网络,实现了对隧道渗水病害的快速检测;最后,利用铁道移动测量系统在武汉采集的数据进行试验验证。试验结果表明,本文提出的改进Mask R-CNN模型的渗水病害检测精度比原算法提升了12%以上,在地铁隧道渗水病害检测中有较好的性能表现。

关键词: 地铁盾构隧道, 渗漏水, 深度学习, 病害检测, Swin transformer

Abstract: Water leakage detection of shield tunnel is the basis of tunnel maintenance. Aiming at the problem that traditional detection methods are difficult to obtain high-quality disease pictures under dim light conditions inside the tunnel, this paper uses lidar to detect internal water leakage in the tunnel. Firstly, based on the characteristics of water leakage disease in the grayscale image of laser point cloud, a data set of water leakage disease in the grayscale image of laser point cloud was established. Then, the Mask R-CNN (Region-CNN) model is used as the benchmark framework, and the Swin transformer network is used as the underlying feature extraction network to realize the rapid detection of tunnel water leakage diseases. Finally, the data collected by the railway mobile measurement system in Wuhan are used for experimental verification. The experimental results show that the detection accuracy of the improved Mask R-CNN model proposed in this paper is higher than that of the original algorithm by more than 12%, and it has good performance in the detection of subway tunnel water leakage diseases.

Key words: metro shield tunnel, leakage, deep learning, disease detection, Swin transformer

中图分类号: