This Title All WIREs
How to cite this WIREs title:
WIREs Data Mining Knowl Discov
Impact Factor: 7.250

Tunnel crack detection using coarse‐to‐fine region localization and edge detection

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

Detecting cracks on the concrete surface is crucial for the tunnel health monitoring and maintenance of Chinese transport facilities, since it is closely related with the structural health and reliability. The automated and efficient tunnel crack detection recently has attracted more research studies, particularly cheap availability of digital cameras makes this issue easier. However, it is still a challenging task due to concrete blebs, stains, and illumination over the concrete surface. This paper presents an efficient crack detection method in the tunnel concrete structure based on digital image processing and deep learning. Three contributions of the paper are summarized as follows. First, we collect and annotate a tunnel crack dataset including three kinds of common cracks that might benefit the research in the field. Second, we propose a new coarse‐to‐fine crack detection method using improved preprocessing, coarse crack region localization and classification, and fine crack edge detection. Third, we introduce a faster region convolutional neural network to develop a coarse crack region localization and classification, then deploy edge extraction to implement the fine crack edge detection, gaining a high‐efficiency and high‐accuracy performance. This article is categorized under: Technologies > Machine Learning Application Areas > Industry Specific Applications Technologies > Classification
The flowchart of our proposed coarse‐to‐fine tunnel crack detection using region localization and edge detection. The tunnel images are collected, annotated, and augmented to get the tunnel crack dataset, then these images are preprocessed by graying, Gaussian filter, and Canny for the following steps. Then, the coarse crack region localization and classification are implemented by introducing the faster region convolutional neural network with region proposal network (RPN), loss function, bounding box regression, and multiclass logistic regression. Finally, the fine crack edge detection is deployed using median filter, direction gradient, and open operator to generate the detailed detection results for the tunnel quality evaluation
[ Normal View | Magnified View ]
The results of fine edge detection. (a) Bounding box region; (b) median filter with adaptive threshold; (c) direction gradient; and (d) open operator with water filling
[ Normal View | Magnified View ]
The PR curve for testing set C and C*
[ Normal View | Magnified View ]
The detection results of noisy images. (a) Normal connection, (b) water leakage, (c) dense cracks
[ Normal View | Magnified View ]
The comparison of left: Original image, middle: Annotation, right: Our detection result
[ Normal View | Magnified View ]
The examples of the crack image dataset. (a) ZXLF: Longitudinal crack (bounding box in blue color), (b) HXLF: Transverse crack (bounding box in red color), (c) XLF: Diagonal crack (bounding box in green color)
[ Normal View | Magnified View ]
Overview of the region proposal network (RPN) in the architecture of faster region convolutional neural network
[ Normal View | Magnified View ]
Image preprocessing: (a) Graying, (b) Gaussian filter, (c) Canny
[ Normal View | Magnified View ]

Browse by Topic

Technologies > Classification
Application Areas > Industry Specific Applications
Technologies > Machine Learning

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts