基于计算机视觉的钢桁架桥梁梁底病害扫描方法  

Disease Scanning Method for Steel Truss Bridge Bottom Based on Computer Vision

作　　者：徐向东 唐志[1] 张剑锋 XU Xiangdong;TANG Zhi;ZHANG Jianfeng

机构地区：[1]贵州省交通规划勘察设计研究院股份有限公司,贵州贵阳550081

出　　处：《上海公路》2024年第3期79-84,134,M0006,共8页Shanghai Highways

基　　金：贵州省交通科技项目(2020-123-023)。

摘　　要：钢桁架桥梁底部由于长期受到自然环境的侵蚀和交通荷载的影响,容易发生锈蚀等现象,若不及时进行检测和维修,可能会导致结构性能下降,甚至发生安全事故。现依托于北盘江大桥,构建了梁底腐蚀典型病害数据集,并基于标准化的图像预处理技术,通过选取合适的卷积神经网络,实现了病害图像的特征自动提取,同时以机器学习的方式,实现了病害分类,完成了对疑似病害区域的检测。The bottom of steel truss bridge is susceptible to corrosion and paint loss due to the long-term erosion of the natural environment and the influence of traffic loads.If not detected and repaired promptly,it may lead to the decline of structural performance and even result in safety accidents.This study,based on the Beipan River Bridge,establishes a dataset of typical corrosion-related defects.Utilizing standardized image preprocessing technology,the automatic feature extraction of disease images is realized by selecting appropriate convolutional neural networks,and the disease classification is realized by machine learning,and the detection of suspected disease areas is completed.

关 键 词：梁底检测 病害识别 深度学习 语义分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U446[自动化与计算机技术—计算机科学与技术]