基于优化的残差网络的锈蚀钢筋图像分类

doi:10.12068/j.issn.1005-3026.2021.11.015

东北大学学报（自然科学版） ›› 2021, Vol. 42 ›› Issue (11): 1625-1633.DOI: 10.12068/j.issn.1005-3026.2021.11.015

基于优化的残差网络的锈蚀钢筋图像分类

方亮^1,2,3，周云^1,2，唐志泉⁴

(1. 湖南大学土木工程学院，湖南长沙410082; 2. 湖南大学工程结构损伤诊断湖南省重点实验室，湖南长沙410082; 3. 湖南农业大学水利与土木工程学院，湖南长沙410128; 4. 湖南大学信息科学与工程学院，湖南长沙410082)

修回日期:2021-04-26 接受日期:2021-04-26 发布日期:2021-11-19
通讯作者: 方亮
作者简介:方亮(1981-)，女，湖南益阳人，湖南大学博士研究生，湖南农业大学讲师；周云(1979-)，男，湖南长沙人，湖南大学教授，博士生导师.
基金资助:
国家自然科学基金资助项目(51878264); 湖南省交通运输厅科技进步与创新项目(201912); 长沙市科技计划项目(kq1801010).

Image Classification of Corroded Steel Reinforcement Based on Optimized Residual Network

FANG Liang^1,2,3， ZHOU Yun^1,2， TANG Zhi-quan⁴

1. College of Civil Engineering， Hunan University， Changsha 410082， China; 2. Hunan Provincial Key Lab on Damage Diagnosis for Engineering Structures， Hunan University， Changsha 410082， China; 3. College of Water Resource ＆ Civil Engineering， Hunan Agricultural University， Changsha 410128， China; 4. College of Computer Science and Electronic Engineering， Hunan University， Changsha 410082， China.

Revised:2021-04-26 Accepted:2021-04-26 Published:2021-11-19
Contact: ZHOU Yun
About author:-
Supported by:
-

摘要/Abstract

摘要： 通过实验研究提出一种基于残差网络(ResNet)的锈蚀钢筋分类方法，为开发锈蚀钢筋现场准确定量评价方法提供新思路和技术参考.以1478根直径12mm和14mm，锈蚀率1.45%~56.10%的钢筋为研究对象，利用工业相机在实验室条件下拍摄图像，结合数据增强技术，共获得23648张样本图像，并根据锈蚀率确定11类标签.基于深度卷积神经网络搭建ResNet结构，并利用Adam算法进行迭代优化，通过对比不同数据集的实验结果评估分类准确率和训练轮数.为验证所提方法的适用性，将不同直径钢筋的样本图像组合成6种数据集进行训练与测试.研究表明，经过100次迭代训练，针对6种数据集的钢筋锈蚀程度分类准确率均在93.2%以上，最高达98.8%.该方法支持混合直径的锈蚀钢筋高精度分类，具有良好的实际应用性.

关键词: 锈蚀钢筋;锈蚀率;深度卷积神经网络;图像分类;残差网络

Abstract: A classification method for corroded steel reinforcements based on the residual network(ResNet) was proposed， and a new scheme and technical reference for further realizing accurate quantitative evaluation on corrosion ratio of steel reinforcement on-site was provided. A total of 1478 steel reinforcements with diameters of 12mm and 14mm and corrosion ratio of 1.45%~56.10% are taken as the experimental objects. The images of steel reinforcements are taken by industrial cameras under laboratory conditions， and the total of 23648 sample images are acquired by using the data enhancement techniques. Then， 11 kinds of sample image labels are determined according to the corrosion ratios. The ResNet structure is built based on the deep convolutional neural network， and the Adam algorithm is used for iterative optimization. The classification accuracy and training rounds are evaluated by comparing the experimental results of different data sets. To verify the applicability of the proposed method， 6 data sets are formed from sample images of different diameter of steel reinforcements for the training and testing. The experimental research shows that after 100 rounds of iterative training， the classification accuracy of corrosion degree of steel reinforcements for 6 kinds of data sets are more than 93.2%， and the highest is 98.8%. It also indicates that the proposed method supports the high-precision classification of corrosion reinforcement with mixed diameters， and has good practical applicability.

Key words: corroded steel reinforcement; corrosion ratio; deep convolutional neural network(CNN); image classification; residual network(ResNet)

中图分类号:

TU375.2

方亮，周云，唐志泉. 基于优化的残差网络的锈蚀钢筋图像分类[J]. 东北大学学报（自然科学版）, 2021, 42(11): 1625-1633.

FANG Liang， ZHOU Yun， TANG Zhi-quan. Image Classification of Corroded Steel Reinforcement Based on Optimized Residual Network[J]. Journal of Northeastern University(Natural Science), 2021, 42(11): 1625-1633.

参考文献

[1]Hou B R，Li X G，Ma X M，et al.The cost of corrosion in China ［J］.Materials Degradation，2017(4):1-10，
[2]童晶，金贤玉，田野，等.基于DIC技术的锈蚀钢筋混凝土表面开裂［J］.浙江大学学报(工学版)，2015，49(2):193-199，217.(Tong Jing，Jin Xian-yu，Tian Ye，et al.Study on surface cracking of corroded reinforced concrete based on DIC method［J］.Journal of Zhejiang University(Engineering Science)，2015，49(2):193-199，217.)
[3]陈海斌，牛荻涛，浦聿修.应用人工神经网络技术评估混凝土中的钢筋锈蚀量［J］.工业建筑，1999，29(2):51-55.(Chen Hai-bin，Niu Di-tao，Pu Yu-xiu.Assessment on corrosive degree of reinforcement in concrete by artificial neural networks［J］.Industrial Construction，1999，29(2):51-55.)
[4]伍明强.建筑混凝土钢筋锈蚀原因及检测方法研究［J］.建材世界，2019，40(1):31-34.(Wu Ming-qiang.Research on reasons and detection methods of corrosion of building concrete reinforced bar［J］.The World of Building Materials，2019，40(1):31-34.)
[5]Hadi S，Rigoberto B.Emerging artificial intelligence methods in structural engineering［J］.Engineering Structures，2018，171:170-189.
[6]冷艳玲，张劲泉，毛燕.混凝土结构钢筋锈蚀率BP神经网络预测模型［J］.西部交通科技，2010(5):5-8，15.(Leng Yan-ling，Zhang Jing-quan，Mao Yan.A model based on BP neural network used to predict corrosion in reinforced concret［J］.Western China Communications Science & Technology，2010(5):5-8，15.)
[7]刘斌云，王鑫，万其微.基于ABC-BP神经网络预测钢筋锈蚀程度［J］.合成材料老化与应用，2019，48(5):54-58.(Liu Bin-yun，Wang Xin，Wan Qi-wei.Predict corrosion degree of steel bars in reinforcing concrete based on ABC-BP neural network［J］.Synthetic Materials Aging and Application，2019，48(5):54-58.)
[8]甘海龙，郭容宽.基于神经网络技术预测锈胀开裂后混凝土中钢筋锈蚀量［J］.科技通报，2019，35(2):144-149.(Gan Hai-long，Guo Rong-kuan.Prediction on corrosive degree of reinforcement concrete by neural network［J］.Bulletin of Science and Technology，2019，35(2):144-149.)
[9]Rawat W，Wang Z H.Deep convolutional neural networks for image classification:a comprehensive review［J］.Neural Computation，2017，29(9):2352-2449.
[10]王洋，刘积仁，赵大哲，等.卷积神经网络在MRI图像诊断中的应用［J］.东北大学学报(自然科学版)，2019，40(2):169-173.(Wang Yang，Liu Ji-ren，Zhao Da-zhe，et al.Application of convolutional neural networks in computer-aided diagnosis of MRI images［J］.Journal of Northeastern University(Natural Science)，2019，40(2):169-173.)
[11]Szegedy C，Liu W，Jia Y，et al.Going deeper with convolutions［C］//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Boston，2015:1-9.
[12]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.Chongqing，2016:770-778.
[13]Cha Y J，Choi W，Suh G，et al.Autonomous structural visual inspection using region-based deep learning for detecting multiple damage types［J］.Computer-Aided Civil and Infrastructure Engineering，2018，33:731-747.
[14]王达磊，彭博，潘玥，等.基于深度神经网络的锈蚀图像分割与定量分析［J］.华南理工大学学报(自然科学版)，2018，46(12):121-127，146.(Wang Da-lei，Peng Bo，Pan Yue，et al.Segmentation and quantitative analysis of corrosion images based on deep neural networks［J］.Journal of South China University of Technology(Natural Science Edition)，2018，46(12):121-127，146.)
[15]中华人民共和国国家标准.混凝土结构设计规范:GB50010—2010［S］.北京:中国建筑工业出版社，2015.(National Standards of the People′s Republic of China.Code for design of concrete structures:GB50010—2010 ［S］.Beijing:China Architecture & Building Press，2015.)
[16]方亮，周云，易督航.锈蚀 HRB500 钢筋混凝土板抗弯性能试验研究［J］.湖南大学学报(自然科学版)，2020，47(11):84-94.(Fang Liang，Zhou Yun，Yi Du-hang.Research on flexural behavior of corroded reinforced concrete slabs with HRB500 bars［J］.Journal of Hunan University(Natural Sciences)，2020，47(11):84-94.)
[17]中华人民共和国国家标准.混凝土结构现场检测技术标准:GB/T50784—2013［S］.北京:中国建筑工业出版社，2013.(National Standards of the People′s Republic of China.Technical standard of in-site inspection of concrete structure:GB/T50784—2013 ［S］.Beijing:China Architecture & Building Press，2013.)
[18]LeCun Y，Bengio Y，Hinton G.Deep learning［J］.Nature，2015，521(7553):436-444.
[19]Zhang R.Making convolutional networks shift-invariant again［J］.arXiv preprint arXiv:1904.11486，2019.
[20]roba L，Grman J，Ravas R.Impact of Gaussian noise and image filtering to detected corner points positions stability［C］//2017 11th International Conference on Measurement.Smolenice:IEEE，2017:123-126.
[21]Murphy K P.Machine learning:a probabilistic perspective［M］.Cambridge:MIT Press，2012.

基于优化的残差网络的锈蚀钢筋图像分类

Image Classification of Corroded Steel Reinforcement Based on Optimized Residual Network

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

编辑推荐

Metrics

本文评价