Research On Image Processing Technology Of Subway Tunnel Crack Based On Feature Analysis

With the rapid development of urban rail transit in China,most of the metro lines are built in underground tunnels,thus their safety is very important for metro transportation.During the construction and use of metro tunnels,cracks inevitably cracks due to temperature,humidity,rock properties and other reasons.Cracks not only affect the stability of tunnels,but also seriously endanger the safety of trains.Therefore,effective crack detection method is very important to provide accurate basic data for the metro tunnel repair and maintenance departments.At present,most of tunnel crack detection methods in our country rely on manual naked eye detection and manual marking,but manual method not only wastes time and energy,but also has strong subjectivity,which is not conducive to tunnel safety detection.Therefore,the use of automated crack detection system which is to replace traditional methods has become a mainstream trend in the development of tunnel engineering non-destructive testing.This thesis firstly uses the Mask homogenization algorithm to balance the overall brightness of the image,equalize the illumination effect,and then,use gray level corrosion to enhance the contrast between the crack and the background.These pre-processing can improve the image quality and reduce the difficulty of subsequent processing.The Gauss-fast median filtering algorithm is used to denoise the graphics,remove the noise of a large number of background components in the image,which is beneficial to subsequent image segmentation and recognition.In the image segmentation process,the image is segmented by using the quadratic improved Otsu algorithm based on the edge information to obtain a binary image,and the obtained binary image is denoised and connected.Then,the skeleton image extraction and burr filtering are performed on the crack image,and the crack is classified and recognized by the projection method and the threshold method.According to the characteristics of the processed binary crack image projected on the coordinate axis,combined with the set threshold,it will be detected.The crack image is divided into lateral,longitudinal and reticular cracks.Finally,various geometric parameters of the crack image are calculated,and the characteristic parameters,such as crack length,width and area,are calculated in the pixel domain according to the crack type.Finally,the actual size of the crack is calculated by actual camera calibration,which provides a scientific and reliable basis for tunnel safety evaluation.The algorithm proposed in this thesis is simulated by using Matlab.Through the detection of several pictures,it is verified that the proposed algorithm has high recognition rate and low miss detection rate.It can effectively extract the geometric eigenvalues of tunnel cracks,thus realizing the automatic identification and detection of cracks.