基于计算机视觉的钢结构表面锈蚀程度检测方法

逯鹏; 赵天淞; 王剑; 常好诵; 郑云; 刘小兰

doi:10.3724/j.gyjzG23062009

基于计算机视觉的钢结构表面锈蚀程度检测方法

doi: 10.3724/j.gyjzG23062009

逯鹏^1,3,
赵天淞²,
王剑²,
常好诵^1,3,
郑云^1,3,
刘小兰²

1. 中冶检测认证有限公司, 北京 100088;
2. 天津城建大学, 天津 300384;
3. 中冶建筑研究总院有限公司, 北京 100088

基金项目:

甘肃省科技计划资助项目(23YFGA0038)。

详细信息

作者简介:
逯鹏,高级工程师,主要从事结构检测鉴定及钢结构耐久性评价等工作。电子信箱:lupeng201501@126.com。

计量
- 文章访问数: 43
- HTML全文浏览量: 2
- PDF下载量: 4
- 被引次数: 0
出版历程
- 收稿日期: 2023-06-20
- 网络出版日期: 2024-09-19

A Method for Detecting Surface Corrosion Degree of Steel Structures Based on Computer Vision

1. Inspection and Certification Co., Ltd., MCC, Beijing 100088, China;
2. Tianjin Chengjian University, Tianjin 300384, China;
3. Central Research Institute of Building and Construction Co., Ltd., MCC Group, Beijing 100088, China

摘要

摘要: 针对钢结构表面锈蚀损伤,传统的人工检测方法耗时、费力,且受检测人员技术水平限制。计算机视觉技术为钢结构表面锈蚀的检测和分类提供了一种快速准确的替代方法。目前常用的锈蚀程度检测技术多基于卷积神经网络结构,由于网络结构自身的缺陷,在进行锈蚀程度分类时存在忽略图像中部分锈蚀特征的问题,导致错误的检测结果。为此,提出了一种基于Vision Transformer网络结构的钢结构表面锈蚀程度识别方法,通过引入自注意力机制(SA)在进行特征提取的过程中保证数据的完整性,并在自建的锈蚀程度图像数据集上进行验证,该方法对锈蚀程度的分类准确率可达到90%。此外,还提出一种基于滑动窗口法的钢结构表面锈蚀程度检测方法,对待检测的钢结构图像进行切割,利用训练好的网络结构进行锈蚀程度检测,将检测后的图像重新拼接,实现钢结构表面锈蚀程度的可视化。
- 钢结构 /
- 表面锈蚀 /
- 损伤检测 /
- 计算机视觉 /
- 滑动窗口法
Abstract: Traditional manual inspection methods for surface corrosion damage in steel structures are time-consuming, labor-intensive, and limited by the technical expertise of the inspection personnel. Computer vision technology provides a fast and accurate alternative method for detecting and classifying the surface corrosion on steel structures. Currently, commonly used methods for corrosion degree detection are based on convolutional neural network (CNN) structures. However, due to inherent flaws in the network structure, there is a problem of neglecting certain rust features in the image during corrosion degree classification, leading to incorrect detection results. A steel structure surface corrosion degree recognition method based on the Vision Transformer network structure was proposed. By introducing self-attention mechanisms (SA) during the feature extraction process, data integrity could be ensured. The proposed method was validated on a self-built dataset of rust severity images, achieving an accuracy ratio of 90% in corrosion degree classification. Furthermore, a steel structure surface corrosion degree detection method based on the sliding-window method was also proposed. This method involves segmenting the steel structure images to be inspected, utilizing a trained network structure for corrosion degree detection, and reassembling the detected images to achieve intelligent detection of corrosion degree on the surface of steel structures.
- steel structure /
- surface corrosion /
- damage detection /
- computer vision /
- sliding-window method

HTML全文

参考文献(22)

[1]	GROSHEK I G, HEBDON M H. Galvanic corrosion of ASTM A1010 steel connected to common bridge steels[J]. Journal of Materials in Civil Engineering, 2020, 32(8), 04020193.
[2]	KERE K J, HUANG Q. Life-cycle cost comparison of corrosion management strategies for steel bridges[J]. Journal of Bridge Engineering, 2019, 24(4),0419007.
[3]	WANG N, ZHAO Q, LI S, et al. Damage classification for masonry historic structures using convolutional neural networks based on still images[J]. Computer-Aided Civil and Infrastructure Engineering, 2018, 33(12): 1073-1089.
[4]	FELICIANO F F, LETA F R, MAINIER F B. Texture digital analysis for corrosion monitoring[J]. Corrosion Science, 2015, 93: 138-147.
[5]	LI Z, LIU Y, HAYWARD R, et al. Knowledge-based power line detection for UAV surveillance and inspection systems[C]//Image & Vision Computing New Zealand, Ivcnz, International Conference. 2009.
[6]	NIKOLIC J, BURRI M, REHDER J, et al. A UAV system for inspection of industrial facilities[C]//Aerospace Conference, 2013 IEEE. 2013.
[7]	BONNÍN-PASCUAL F, ORTIZ A. Detection of cracks and corrosion for automated vessels visual inspection[C]//Proceedings of the 13th International Conference of the Catalan Association for Artificial Intelligence. l’Espluga de Francolí: Tarragona, 2010:20-22.
[8]	BENTO M P, DE MEDEIROS F N, PAULA JR I C, et al. Image processing techniques applied for corrosion damage analysis[C]//Proceedings of the XXII Brazilian Symposium on Computer Graphics and Image Processing. Rio de Janeiro: 2009.
[9]	AHUJA S K, SHUKLA M K. A survey of computer vision based corrosion detection approaches[C]//Information and Communication Technology for Intelligent Systems (ICTIS 2017). 2018:55-63.
[10]	MEDEIROS F N S, RAMALHO G L B, BENTO M P, et al. On the evaluation of texture and color features for nondestructive corrosion detection[J/OL]. EURASIP Journal on Advances in Signal Processing, 2010[2010-07-07].https://doi.org/DOI: 10.1155/2010/817473.
[11]	CHEN P H, SHEN H K, LEI C Y, et al. Fourier-transform-based method for automated steel bridge coating defect recognition[J]. Procedia Engineering, 2011, 14: 470-476.
[12]	SHEN H K, CHEN P H, CHANG L M. Automated steel bridge coating rust defect recognition method based on color and texture feature[J]. Automation in Construction, 2013, 31(5): 338-356.
[13]	KRIZHEVSKY A, SUTSKEVER I, HINTON G. ImageNet classification with deep convolutional neural networks[J]. Advances In Neural Information Processing Systems, 2012, 25(2),3065386.
[14]	ATHA D J, JAHANSHAHI M R. Evaluation of deep learning approaches based on convolutional neural networks for corrosion detection[J]. Structural Health Monitoring, 2018,17(5):1110-1128.
[15]	WANG Y, SHEN X, WU K, et al. Corrosion grade recognition for weathering steel plate based on a convolutional neural network[J]. Measurement Science & Technology, 2022,33(9),095014.
[16]	ZHANG S, DENG X, LU Y, et al. A channel attention based deep neural network for automatic metallic corrosion detection[J]. Journal of Building Engineering, 2021, 42, 103046.
[17]	XU J, GUI C, HAN Q. Recognition of rust grade and rust ratio of steel structures based on ensembled convolutional neural network[J]. Computer-Aided Civil and Infrastructure Engineering, 2020, 35(10): 1160-1174.
[18]	DOSOVITSKIY A, BEYER L, KOLESNIKOV A,et al.An image is worth 16×16 words: transformers for image recognition at scale[C]//International Conference on Learning Representations.2021.
[19]	中国国家标准化管理委员会.涂覆涂料前钢材表面处理表面清洁度的目视评定第1部分:未涂覆过的钢材表面和全面清除原有涂层后的钢材:GB/T 8923.1—2011[S].北京:中国标准出版社,2011.
[20]	SIMONYAN K, ZISSERMAN A.Very deep convolutional Networks for large-scale image recognition[J/OL].Computer Science, 2014[2014-09-04].https://doi.org/10.48550/arXiv.1409.1556.
[21]	SANDLER M, HOWARD A, ZHU M, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).IEEE, 2018.
[22]	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016: 770-778.