Abstract

Most of the tunnel projects are related to the national economy and people’s livelihood, and their operational safety is of paramount importance. Tunnel safety accidents or hidden safety hazards often start from subtleties. Therefore, the identification of tunnel cracks is a key part of tunnel safety control. The development of computer vision technology has made it possible for the automatic detection of tunnel cracks. Aiming at the problem of low recognition accuracy of existing crack recognition algorithms, this paper uses an improved homomorphic filtering algorithm to dehaze and clear the collected images according to the characteristics of tunnel images and uses an adaptive median filter to denoise the grayscaled image. The extended difference of Gaussian function is used for edge extraction, and the morphological opening and closing operations are used to remove noise. The breakpoints of the binary image are connected after removing the noise to make the image more in line with the actual situation. Aiming at the identification of tunnel crack types, the block index is proposed and used to distinguish linear cracks and network cracks. Using the histogram-like method to distinguish linear cracks in different directions can well solve the mixed crack situation in an image. Compared with the traditional method, the recognition rate of the new algorithm is increased to 94.5%, which is much higher than the traditional crack recognition algorithm. The average processing time of an image is 5.2 s which is moderate, and the crack type discrimination accuracy is more than 92%. In general, the new algorithm has good prospects for theoretical promotion and high engineering application value.

Highlights

  • Tunnel engineering has complex forces and its structural safety is of great importance

  • E image processing methods of tunnel lining cracks can be divided into two categories: one is the deep learning method based on convolutional neural network and the other is the grayscale image processing method. e method of deep learning has a high accuracy for tunnel lining crack detection

  • It takes a lot of time to establish and train the model because of the need of disease annotation and a large number of images. e traditional grayscale image processing method is to determine the location of the cracks by the sudden change of the gray value

Read more

Summary

Introduction

Tunnel engineering has complex forces and its structural safety is of great importance Diseases such as cracks, water leakage, and lining deformation often occur in operating tunnels. Combining computer vision and image processing technology to automatically detect tunnel lining cracks has become the main research direction. E method of deep learning has a high accuracy for tunnel lining crack detection. The surface of the tunnel lining has oil stains and water leakage, and the image is affected by noise, which will affect the Mathematical Problems in Engineering accuracy of crack recognition. E new algorithm has better processing results for crack images that contain a lot of complex noise, uneven illumination, and uneven contrast. Erefore, this paper proposes a computer vision-based inspection method for surface cracks in tunnel linings.

Image Preprocessing
Crack Direction Classification
Conclusion
Findings
Disclosure
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call