Research On Surface Crack Detection Method Of Railway Tunnel Image

China is in a stage of large-scale railway construction and deployment.Tunnel engineering,as a critical discipline of rail transportation,has also experienced rapid development in recent years.The safety of tunnels is essential for railway transportation.Therefore,railway tunnels will face regular maintenance and inspection in the foreseeable future.Cracks are one of the most common and severe surface defects in railway tunnels.If left untreated,they may lead to other problems such as structural detachment and water seepage,which can affect the safe operation of the railway.In the complex environment of tunnels,manual inspection methods are inefficient,have low detection accuracy,and consume manpower and resources,making it difficult to meet the demand.Based on the advantages of digital image processing methods,such as fast detection speed and high detection accuracy,this paper studied the detection method of crack defects in railway tunnels.The main work of this paper is as follows:(1)This paper summarizes the methods of detecting crack defects in tunnels both domestically and internationally,as well as the main problems encountered during the process of detecting crack defects in tunnels.The paper also analyzes and summarizes the commonly used algorithms in the image preprocessing,image segmentation,and feature extraction stages of the crack detection process.(2)In response to the problems of uneven grayscale distribution,low contrast,lack of prominent image details,and concrete pitting and unevenness on the surface of tunnel linings,proposing a method for rough extraction of cracks.The method consists of two aspects: first,the crack image preprocessing process based on Retinex equalization and limited linear stretching;second,the adaptive threshold segmentation process based on block division.The image preprocessing part uses Retinex equalization and limited linear stretching to equalize the illumination and improve the contrast of the crack image.The adaptive threshold segmentation method under image block division first divides the image into a certain number of blocks according to the tunnel crack characteristics.Then,based on the grayscale characteristics of each region block,the image is divided into background,disease,and other areas.Finally,different threshold methods are used for different regions to obtain the roughly segmented crack image.(3)To address the problem of difficult crack extraction due to various types of cracks and complex interference in railway tunnel images,a crack fine detection process based on structural feature analysis is designed.The component features in the segmented image are analyzed,and different extraction algorithms are selected for each component,including point components based on zero-order moment,block components based on rectangularity,line components based on cumulative probability Hough transform,and pseudo-crack components based on skeleton features.The crack skeleton extraction algorithm is improved,which can effectively remove the influence of redundant pixels.Finally,the crack width is calculated based on the mean and standard deviation of the pixel neighborhood of the crack skeleton.The experimental results show that the detection rate of traditional crack image and tunnel crack image can reach 92.1% and 86% respectively.