Efficient Crack Detection Method for Tunnel Lining Surface Cracks Based on Infrared Images

Abstract

The detection of tunnel lining cracks is a very key procedure in the inspection of tunnels. Traditional image-processing approaches are commonly based on the characteristic that the grayscale value of the crack is a local minimum. However, issues such as low contrast, uneven illumination, and severe noise pollution generally exist in a tunnel lining image. Hence, the traditional image-processing method cannot effectively detect cracks on the tunnel lining surface. This paper presents a three-step method to identify and extract cracks from infrared images of tunnel lining. First, the image is preprocessed in the frequency domain. Second, the conditional texture anisotropy of each pixel is computed in an image subblock, and the optimum threshold is obtained with an iteration method. Thus, the cracks in the image subblock are determined according to the threshold. Finally, the cracks in each subregion are connected. Experimental results show that the proposed method can effectively detect tunnel lining surface cracks.