基于改进DeepLabV3+和渗流算法的隧道裂缝检测研究

Research on tunnel crack detection based on improved DeepLabV3+ and percolation algorithm

摘要

针对隧道衬砌裂缝检测中深度学习边缘定位不准、传统渗流算法效率低的问题,提出一种“深度学习粗分割+渗流算法精提取”两阶段方法。第一阶段采用改进的DeepLabV3+模型(轻量化主干网络融合CBAM注意力、优化ASPP模块及Dice损失)实现高召回率预分割;第二阶段基于裂缝骨架引导渗流生长,并结合形态学特征约束,实现边缘精细化提取与宽度测量。本研究自主构建了包含20504张像素级标注图像的隧道裂缝数据集,图像涵盖拱顶、拱腰、边墙等不同部位及干燥、潮湿等多种表面状态。基于数据集试验结果表明,预分割模块准确率达90.1%、召回率86.7%;改进的渗流算法在保持高召回率的同时将精确率提升至98.5%,计算效率提高约20倍。工程验证表明,算法对0.1 mm以上裂缝的检出率达84%以上,宽度测量平均绝对误差小于0.3 mm。该方法有效兼顾了检测精度与计算效率,为隧道衬砌裂缝自动化检测提供了可行技术方案。

Abstract

To address the problem of poor edge localization in deep learning-based methods and low efficiency of traditional percolation algorithms for tunnel crack detection, a two-stage approach that integrates an improved DeepLabV3+ with a skeleton-guided percolation algorithm was proposed. In the first stage, an improved DeepLabV3+ model with a lightweight backbone network, incorporating a CBAM attention module, an optimized ASPP module, and a Dice loss function, was developed to achieve high-recall crack pre-segmentation. In the second stage, a skeleton-guided percolation growth strategy combined with morphological constraints was applied to refine crack edges and measure crack widths. A tunnel crack dataset containing 20504 pixel-level annotated images was constructed, covering various tunnel lining regions (crown, haunch, sidewall) and surface conditions (dry, wet, stained). Experimental results on this dataset showed that the pre-segmentation module achieved an accuracy of 90.1% and a recall of 86.7%. The improved percolation algorithm increased the precision to 98.5% while maintaining high recall, and improved computational efficiency by approximately 20 times. Engineering validation demonstrated a detection rate exceeding 84% for cracks wider than 0.1 mm, with a mean absolute error of less than 0.3 mm. The proposed method effectively balanced detection accuracy and computational efficiency, providing a feasible solution for automated tunnel lining crack detection.

关键词

隧道病害 / 裂缝检测 / 深度学习 / 渗流算法 / 图像处理

Key words

tunnel defect / crack detection / deep learning / percolation algorithm / image processing

参考文献

[1] 中华人民共和国交通运输部 . 2020年交通运输行业发展统计公报[EB/OL]. (2021—05—19)[2026—01—15]. https://xxgk.mot.gov.cn/2020/jigou/zhghs/202105/t20210517_3593412.html

[2] 李固华, 郭建国. 隧道衬砌裂缝和渗漏的成因、预防及治理[J]. 铁道建筑, 2003, 43(1): 23-25.

[3] LI Guhua, GUO Jianguo. The causes inducing crack and water leakage in tunnel lining and its prevention and treatment[J]. Railway Engineering, 2003, 43(1): 23-25.

[4] 张素磊, 张顶立, 刘昌. 公路运营隧道衬砌裂缝长期监测及分析[J]. 现代隧道技术, 2017, 54(3): 17-25.

[5] ZHANG Sulei, ZHANG Dingli, LIU Chang. Long—term monitoring and analysis of lining cracks in operating highway tunnels[J]. Modern Tunnelling Technology, 2017, 54(3): 17-25.

[6] 陈志敏, 师浩博, 张润龙, 等. 基于裂缝形状与特征的衬砌结构开裂截面稳定性评估[J]. 隧道与地下工程灾害防治, 2026, 8(1): 13-21.

[7] CHEN Zhimin, SHI Haobo, ZHANG Runlong, et al. Stability evaluation of cracked section of lining structure based on crack shape and characteristics[J]. Hazard Control in Tunnelling and Underground Engineering, 2026, 8(1): 13-21.

[8] 陈湘生, 徐志豪, 包小华, 等. 隧道病害监测检测技术研究现状概述[J]. 隧道与地下工程灾害防治, 2020, 2(3): 1-12.

[9] CHEN Xiangsheng, XU Zhihao, BAO Xiaohua, et al. Overview of research on tunnel defects monitoring and detection technology[J]. Hazard Control in Tunnelling and Underground Engineering, 2020, 2(3): 1-12.

[10] XING Y D, SHI Y X, JIAO X J, et al. Research on software and algorithms for displacement monitoring system based on machine vision[C]// 2025 8th International Conference on Advanced Algorithms and Control Engineering (ICAACE). Shanghai: IEEE, 2025: 1691-1699.

[11] 宋益, 赵宁雨, 颜畅, 等. 隧道衬砌裂缝实时分割的Mobile—PSPNet方法[J]. 铁道科学与工程学报, 2022, 19(12): 3746-3757.

[12] SONG Yi, ZHAO Ningyu, YAN Chang, et al. The Mobile—PSPNet method for real—time segmentation of tunnel lining cracks[J]. Journal of Railway Science and Engineering, 2022, 19(12): 3746-3757.

[13] 周中, 闫龙宾, 张俊杰, 等. 基于YOLOX—G算法的隧道裂缝实时检测[J]. 铁道科学与工程学报, 2023, 20(7): 2751-2762.

[14] ZHOU Zhong, YAN Longbin, ZHANG Junjie, et al. Real—time detection of tunnel cracks based on YOLOX—G algorithm[J]. Journal of Railway Science and Engineering, 2023, 20(7): 2751-2762.

[15] 鲍艳, 梅崇斌, 徐鹏宇, 等. 基于改进YOLOv8算法的地铁隧道裂缝识别方法研究[J]. 隧道建设(中英文), 2024, 44(10): 1961-1970.

[16] BAO Yan, MEI Chongbin, XU Pengyu, et al. Crack identification in metro tunnels based on improved YOLOv8 algorithm[J]. Tunnel Construction, 2024, 44(10): 1961-1970.

[17] 刘毅, 陈一丹, 高琳, 等. 基于多尺度特征融合的轻量化道路提取模型[J]. 浙江大学学报(工学版), 2024, 58(5): 951-959.

[18] LIU Yi, CHEN Yidan, GAO Lin, et al. Lightweight road extraction model based on multi—scale feature fusion[J]. Journal of Zhejiang University (Engineering Science), 2024, 58(5): 951-959.

[19] 孟诗乔, 张啸天, 乔甦阳, 等. 基于深度学习的网格优化裂缝检测模型研究[J]. 建筑结构学报, 2020, 41(增刊2): 404-410.

[20] MENG Shiqiao, ZHANG Xiaotian, QIAO Suyang, et al. Research on grid optimized crack detection model based on deep learning[J]. Journal of Building Structures, 2020, 41(Suppl.2): 404-410.

[21] 黄思文, 包腾飞, 李扬涛, 等. 基于改进DeeplabV3+网络的水工混凝土裂缝语义分割方法[J]. 水利水电科技进展, 2023, 43(1): 81-86.

[22] HUANG Siwen, BAO Tengfei, LI Yangtao, et al. Semantic segmentation method of hydraulic concrete cracks based on improved DeeplabV3+ network[J]. Advances in Science and Technology of Water Resources, 2023, 43(1): 81-86.

[23] CHEN H W, WANG L, ZHANG L, et al. Research on land cover type classification method based on improved MaskFormer for remote sensing images[J]. PeerJ Computer Science, 2023, 9: e1222.

[24] 刘岗顶, 王玲军, 张玉常, 等. 基于改良MaskFormer的鱼眼相机天空图像分割方法[J]. 科学技术创新, 2025(16): 223-228.

[25] LIU Gangding, WANG Lingjun, ZHANG Yuchang, et al. Improved MaskFormer for sky image segmentation using fisheye cameras[J]. Scientific and Technological Innovation Information, 2025(16): 223-228.

[26] GUO Y F, LUO W, DING W Y, et al. EAPEM: an edge—aware—prototype—based efficient MaskFormer for identifying rock mass structure of cantilever roadheader tunnel face[J]. Applications in Engineering Science, 2025, 24: 100274.

[27] 蒋仕新, 唐椿程, 杨建喜, 等. 基于改进Segformer的混凝土桥梁表观病害轻量级识别方法[J]. 中国公路学报, 2024, 37(2): 77-87.

[28] JIANG Shixin, TANG Chuncheng, YANG Jianxi, et al. Lightweight detection method of surface damage on concrete bridges based on improved segformer[J]. China Journal of Highway and Transport, 2024, 37(2): 77-87.

[29] 龚玉磊, 张亚丽, 章红梅, 等. 混凝土构件表面裂缝分割的改进SegFormer方法[J]. 结构工程师, 2025, 41(4): 31-40.

[30] GONG Yulei, ZHANG Yali, ZHANG Hongmei, et al. Improved SegFormer model for surface crack segmentation in concrete components[J]. Structural Engineers, 2025, 41(4): 31-40.

[31] 周勇军, 罗楠, 孙延晨, 等. 混凝土桥梁整体表观多缺陷图像精细分割方法[J]. 哈尔滨工业大学学报, 2025, 57(6): 103-115.

[32] ZHOU Yongjun, LUO Nan, SUN Yanchen, et al. Fine—grained image segmentation method for holistic surface multi—defects in concrete bridges[J]. Journal of Harbin Institute of Technology, 2025, 57(6): 103-115.

[33] YAMAGUCHI T. Image processing based on percolation model[J]. IEICE Transactions on Information and Systems, 2006, E89—D(7): 2044-2052.

[34] YAMAGUCHI T, HASHIMOTO S. Automated crack detection for concrete surface image using percolation model and edge information[C]// IECON 2006—32nd Annual Conference on IEEE Industrial Electronics. Paris, France: IEEE, 2007: 3355-3360.

[35] 瞿中, 郭阳, 鞠芳蓉. 一种基于改进渗流模型的混凝土表面裂缝快速检测算法[J]. 计算机科学, 2017, 44(1): 300-302.

[36] QU Zhong, GUO Yang, JU Fangrong. Algorithm of accelerated cracks detection based on improved percolation model in concrete surface image[J]. Computer Science, 2017, 44(1): 300-302.

[37] 安世全, 曹悦欣, 瞿中. 多因子判定与渗流模型相结合的裂缝检测算法[J]. 计算机应用, 2019, 39(1): 281-286.

[38] AN Shiquan, CAO Yuexin, Quzhong. Crack detection algorithm based on multi—factor decision and percolation model[J]. Journal of Computer Applications, 2019, 39(1): 281-286.

[39] ZHANG T Y, SUEN C Y. A fast parallel algorithm for thinning digital patterns[J]. Communications of the ACM, 1984, 27(3): 236-239.

[40] 彭妍, 郭君斌, 于传强, 等. 基于平面变换的高精度相机标定方法[J]. 北京航空航天大学学报, 2022, 48(7): 1297-1303.

[41] PENG Yan, GUO Junbin, YU Chuanqiang, et al. Calibration method for high precision camera based on plane transformation[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(7): 1297-1303.

[42] 王晓欣. 基于深度学习的复杂环境下混凝土桥梁裂缝智能识别技术研究[D]. 大连: 大连理工大学, 2024: 43-47.

[43] WANG Xiaoxin. Research on intelligent recognition technology of concrete bridge cracks in complex environment based on deep learning[D]. Dalian: Dalian University of Technology, 2024: 43-47.

基金资助

福厦泉国家自主创新示范区协同创新平台资助项目(2024-P-006)

福建省第八批省引才“百人计划”创新创业资助项目()