A new image-based method for concrete bridge bottom crack detection

Crack detection is crucial for safety and cost-effective maintenance of concrete structures. Researchers have proposed several methods based on machine vision techniques to inspect the cracks on the bottom surface of concrete bridges, such as Fujita's method. However, it is difficult to obtain high-quality images and image processing results because of complex environmental and light conditions under bridges. In this study, we propose a new method of crack image processing for concrete bridge bottom crack inspections to solve this problem. We build a machine vision system based on this method, which could detect cracks in real time. We examine the efficiency of the proposed system by evaluating it with real images of cracks and compare them with other image processing methods. In terms of efficiency and accuracy of detecting cracks, experimental results show that proposed method is superior to conventional methods in complex environments under bridges.