基于融合注意力机制DAUNet的最大裂缝宽度计算

doi:10.16183/j.cnki.jsjtu.2023.593

摘要/Abstract

摘要：

裂缝是最常见的路面病害之一,会影响到道路行车安全.针对人工排查和测定路面裂缝成本高、耗时长等问题,提出了一种基于视觉传感智能检测路面裂缝最大缝宽的方法.该方法优化了DAUNet网络,融合了注意力机制,提高了裂缝图像分割的精度;然后将分割出来的裂缝通过腐蚀迭代、连通域判别及象限划分等处理,可以更加准确计算出不同走向裂缝的最大缝宽.实验结果表明:优化后的DAUNet在数据集中所有图像的最佳测量值的平均值(sOIS)方面提高了3.15%,计算最大缝宽的精度相比于目前较优的最大缝宽计算方法提高了3.09个百分点,时间缩短了89.06%.

关键词: 裂缝, 图像分割, 注意力机制, 最大缝宽计算

Abstract:

Cracks are one of the most common pavement diseases, which will affect road traffic safety. To address the high cost and time-consuming challenges associated with manual investigation and determination of pavement cracks, a method based on image processing is proposed to intelligently detect the maximum crack width. The DAUNet framework is optimized, the attention mechanism is integrated, and the accuracy of crack segmentation is improved. Then, the segmented cracks are processed through corrosion iteration, connected domain discrimination, and quadrant division, so that the maximum width of cracks with different directions can be calculated more accurately. Experimental results show that the optimized DAUNet has improved the evaluation index sOIS by 3.15%, increased the accuracy of calculating the maximum crack width by 3.09 percentage points in comparison with the current optimal maximum crack width calculation method, and shortened the time by 89.06%.

Key words: crack, image segmentation, attention mechanism, maximum crack width calculation

中图分类号:

汪维, 阮雅端, 顾鹏, 陈启美. 基于融合注意力机制DAUNet的最大裂缝宽度计算[J]. 上海交通大学学报, 2025, 59(12): 1866-1877.

WANG Wei, RUAN Yaduan, GU Peng, CHEN Qimei. Calculation of Maximum Crack Width Based on DAUNet Integrating Attention Mechanism[J]. Journal of Shanghai Jiao Tong University, 2025, 59(12): 1866-1877.

图/表 22

图1

图2

图3

图4

图5

图6

图7

图8

图9

图10

图11

图12

表1

图13

图14

图15

表2

图16

表3

图17

表4

图18

参考文献 20

[1]	刘宇飞, 樊健生, 聂建国, 等. 结构表面裂缝数字图像法识别研究综述与前景展望[J]. 土木工程学报, 2021, 54(6): 79-98.
	LIU Yufei, FAN Jiansheng, NIE Jianguo, et al. Review and prospect of digital-image-based crack detection of structure surface[J]. China Civil Engineering Journal, 2021, 54(6): 79-98.
[2]	刘凡, 王君锋, 陈峙宇, 等. 基于并行注意力UNet的裂缝检测方法[J]. 计算机研究与发展, 2021, 58(8): 1718-1726.
	LIU Fan, WANG Junfeng, CHEN Zhiyu, et al. Crack detection method based on parallel attention UNet[J]. Computer Research and Development, 2021, 58(8): 1718-1726.
[3]	饶勇成, 韩晓健, 肖飞, 等. 基于深度学习的混凝土结构多病害检测[J]. 建筑结构, 2021, 51(Sup.2): 1439-1445.
	RAO Yongcheng, HAN Xiaojian, XIAO Fei, et al. Multiple disease detection of concrete structures based on deep learning[J]. Building Structure, 2021, 51 (Sup.2): 1439-1445.
[4]	JI A K, XUE X L, WANG Y N, et al. An integrated approach to automatic pixel-level crack detection and quantification of asphalt pavement[J]. Automation in Construction, 2020, 114: 103176. doi: 10.1016/j.autcon.2020.103176 URL
[5]	徐鹏, 祝轩, 姚丁, 等. 沥青路面养护智能检测与决策综述[J]. 中南大学学报(自然科学版), 2021, 52(7): 2099-2117.
	XU Peng, ZHU Xuan, YAO Ding, et al. Review on intelligent detection and decision-making of asphalt pavement maintenance[J]. Journal of Central South University: Science and Technology, 2021, 52(7): 2099-2117.
[6]	陈娟, 张彦铎, 卢涛. 基于像素差分卷积神经网络的类岩石裂缝检测方法[J]. 武汉工程大学学报, 2023, 45(1): 81-86.
	CHEN Juan, ZHANG Yanduo, LU Tao. Rock-like crack detection via pixel differential convolution neural network[J]. Journal of Wuhan Institute of Technology, 2023, 45(1): 81-86.
[7]	王宝坤, 王如路, 陈锦剑, 等. 基于深度学习的盾构隧道表观病害自动检测方法[J]. 上海交通大学学报. 2024, 58(11): 1716-1723. doi: 10.16183/j.cnki.jsjtu.2023.089
	WANG Baokun, WANG Rulu, CHENG Jinjian, et al. Automatic detection method for surface diseases of shield tunnel based on deep learning[J]. Journal of Shanghai Jiao Tong University, 2024, 58(11): 1716-1723.
[8]	徐志刚, 车艳丽, 李金龙, 等. 路面破损图像自动处理技术研究进展[J]. 交通运输工程学报, 2019, 19(1): 172-190.
	XU Zhigang, CHE Yanli, LI Jinlong, et al. Research progress on automatic image processing technology for pavement distress[J]. Journal of Traffic and Transportation Engineering, 2019, 19(1): 172-190.
[9]	田萱, 王亮, 丁琪. 基于深度学习的图像语义分割方法综述[J]. 软件学报, 2019, 30(2): 440-468.
	TIAN Xuan, WANG Liang, DING Qi. Overview of image semantic segmentation methods based on deep learning[J]. Journal of Software, 2019, 30(2): 440-468.
[10]	JIN S, LEE S E, HONG J W. A vision-based approach for autonomous crack width measurement with flexible kernel[J]. Automationin Construction, 2020, 110: 103019.
[11]	徐志刚, 赵祥模, 宋焕生, 等. 基于直方图估计和形状分析的沥青路面裂缝识别算法[J]. 仪器仪表学报, 2010, 31(10): 2260-2266.
	XU Zhigang, ZHAO Xiangmo, SONG Huansheng, et al. Asphalt pavement crack recognition algorithm based on histogram estimation and shape analysis[J]. Chinese Journal of Scientific Instrument, 2010, 31(10): 2260-2266.
[12]	卫军, 曹龙飞, 曾艳霞, 等. 混凝土结构裂缝特征提取研究[J]. 武汉理工大学学报, 2011, 33(11): 72-75.
	WEI Jun, CAO Longfei, ZENG Yanxia, et al. Study on feature extraction of concrete structure cracks[J]. Journal of Wuhan University of Technology, 2011, 33(11): 72-75.
[13]	瞿中, 林丽丹, 郭阳, 等. 形态学与区域延伸相结合的图像裂缝检测算法研究[J]. 计算机科学, 2014, 41(11): 297-300. doi: 10.11896/j.issn.1002-137X.2014.11.058
	QU Zhong, LIN Lidan, GUO Yang, et al. Algorithm of image crack detection based on morphology and region extends[J]. Computer Science, 2014, 41(11): 297-300. doi: 10.11896/j.issn.1002-137X.2014.11.058
[14]	YANG F, ZHANG L, YU S, et al. Feature pyramid and hierarchical boosting network for pavement crack detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(4): 1525-1535. doi: 10.1109/TITS.6979 URL
[15]	FAN Z, LI C, CHEN Y, et al. Automatic crack detection on road pavements using encoder-decoder architecture[J]. Materials (Basel), 2020, 13(13): 2960. doi: 10.3390/ma13132960 URL
[16]	POLOVNIKOV V, ALEKSEEV D, VINOGRADOV I, et al. DAUNet: Deep augmented neural network for pavement crack segmentation[J]. IEEE Access, 2021, 9: 125714-125723. doi: 10.1109/ACCESS.2021.3111223 URL
[17]	王中宇, 李萌, 葛乐矣. 不规则裂纹宽度的一种分段评定法[J]. 航空动力学报, 2008, 23(3): 552-556.
	WANG Zhongyu, LI Meng, GE Leyi. Subsection method to evaluate width of irregular crack[J]. Journal of Aerospace Power, 2008, 23(3): 552-556.
[18]	江涛锋, 唐永圣, 陆志溢, 等. 基于语义分割的混凝土裂缝图像识别方法研究[J]. 工程勘察, 2023, 51(4): 42-47.
	JIANG Taofeng, TANG Yongsheng, LU Zhiyi, et al. Research on method of concrete crack recognition based on semantic segmentation of images[J]. Geotechnical Investigation & Surveying, 2023, 51(4): 42-47.
[19]	余加勇, 刘宝麟, 尹东, 等. 基于YOLOv5和U-Net3+的桥梁裂缝智能识别与测量[J]. 湖南大学学报(自然科学版), 2023, 50(5): 65-73.
	YU Jiayong, LIU Baolin, YIN Dong, et al. Intelligent identification and measurement of bridge cracks based on YOLOv5 and U-Net3+[J]. Journal of Hunan University (Natural Sciences), 2023, 50(5): 65-73.
[20]	阮雅端, 汪维, 顾鹏, 等. 一种适用于路面裂缝的最大裂缝宽度测定方法:CN116012348A[P]. 2023-04-25[2023-11-19].
	RUAN Yaduan, WANG Wei, GU Peng, et al. A method for measuring the maximum crack width of road surface cracks: CN116012348A[P]. 2023-04-25[2023-11-19].

硬件名称	硬件配置
中央处理器(CPU)	Intel(R) Core(TM) i7-7700
内存	8 GB
显卡	GeForce GTX 1080 Ti
显存	12 GB

方法	sODS	sOIS
FPHBN	0.647	0.591
DAUNet	0.750	0.731
DAUNet+SE	0.751	0.757
DAUNet+CBAM	0.755	0.754

图片	迭代次数	时间/s	像素宽度
图16(a)	1	0.070	40.79
	2	0.040	40.79
	4	0.048	227.98
	6	0.020	40.79
图16(b)	1	0.049	33.24
	2	0.030	35.00
	4	0.021	35.00
	6	0.016	35.00

方法	平均相对误差/%	时间/s
八方向搜索法	26.34	0.64
最大内切圆法	10.22	1.89
本文方法	7.13	0.07