Building Crack Identification Based on Convolutional Neural Network and Regional Growth Method

被引：0
|
作者：
Wu Z. [1 ,2 ]
Jia D. [1 ,2 ]
Wang Q. [1 ,2 ]
机构：
[1] School of Mechanics, Civil Engineering and Architecture, Northwestern Polytechnical University, Xi'an
[2] School of Mechanics and Civil Engineering, China University of Mining and Technology, Xuzhou
来源：
Yingyong Jichu yu Gongcheng Kexue Xuebao/Journal of Basic Science and Engineering | 2022年 / 30卷 / 02期
关键词：
Convolutional neural network; Crack identification; Pixel-level crack; Precision test; Regional growth method;
D O I：
10.16058/j.issn.1005-0930.2022.02.006
中图分类号：
学科分类号：
摘要：
The results of building crack recognition based on convolutional neural network(CNN) are bin images with crack, not the crack itself. A two-stage approach, which combined CNN and regional growth method, is proposed to extract a pixel-level crack feature. Data expansion method is used to establish crack image database. Five convolutional neural networks, Alexnet, Vgg16, Vgg19, Inception-V3 and ResNet50 are selected for crack identification. Comprehensively considered the test accuracy of image samples, the crack recognition accuracy of the single image and the confidence of background image, the optimal CNN are selected for crack identification, and the bin images with crack patch are obtained. The regional growth method is used to extract the crack feature from the crack images identified by CNN, and the pixel-level crack images are acquired. The research shows that the Inception-V3 network has higher recognition accuracy. By using the regional growth method to extract the crack feature, a high-precision pixel-level crack characteristic image can be obtained. The research provides a high-precision method for identifying building crack. © 2022, The Editorial Board of Journal of Basic Science and Engineering. All right reserved.
引用
收藏
页码：317 / 327
页数：10
相关论文
共 19 条
  • [1] Xu Y, Wei S Y, Bao Y Q, Et al., Automatic seismic damage identification of reinforced concrete columns form images by a region-based deep convolution neural network, Structural Control & Health Monitoring, 26, 3, (2019)
  • [2] Xue Yadong, Li Yicheng, A method of disease recognition for shield tunnel lining based on deep learning, Journal of Hunan University: Natural Sciences, 45, 3, pp. 100-109, (2018)
  • [3] Modarres C, Asrorga N, Droguett E L, Et al., Convolutional neural networks for automated damage recognition and damage type identification, Structural Control & Health Monitoring, 25, 10, (2018)
  • [4] Liu Xingen, Chen Yingying, Zhu Aixi, Et al., Tunnel crack identification based on deep learning, Journal of Guangxi University: Natural Science Edition, 43, 6, pp. 2243-2251, (2018)
  • [5] Cha Y J, Choi W., Deep learning-based crack damage detection using convolution neural networks, Computer Aided in Civil and Infrastructure Engineering, 32, 5, pp. 361-378, (2017)
  • [6] Li Liangfu, Ma Weifei, Li Li, Et al., Research on detection algorithm for bridge cracks based on deep learning, Acta Automatica Sinica, 44, 3, pp. 1-16, (2018)
  • [7] Zhang L Y, Zhou G C, Han Y, Et al., Application of internet of things technology and convolutional neural network model in bridge crack detection, IEEE Access, 32, pp. 39442-39451, (2018)
  • [8] Wu R T, Jahanshahi M R., Deep convolutional neural network for structural dynamic response estimation and system identification, Journal of Engineering Mechanics, 145, 1, (2019)
  • [9] Sha Aimin, Tong Zheng, Gao Jie, Recognition and measurement of pavement disasters based on convolutional neural network, China Journal of Highway and Transport, 31, 1, pp. 1-10, (2018)
  • [10] Dung C V, Anh L D., Autonomous concrete crack detection using deep fully convolutional neural network, Automation in Construction, 99, pp. 52-58, (2019)
12