| With the construction and development of road tunnels in my country,the number of tunnel mileage is increasing rapidly,and the maintenance of road tunnels becomes more and more important.A non-contact rapid detection technology for detecting tunnel diseases is also constantly being developed.For the vehicle-mounted tunnel lining crack detection system,the camera is affected by the vibration of the vehicle body,and the output image sequence will produce errors,which will affect the tunnel lining crack detection work.Aiming at the vehiclemounted tunnel lining crack detection system,this paper completes the system scheme design of the crack detection system,and proposes a vehicle-mounted tunnel lining crack detection method based on image compensation.The main contents of the paper are as follows:Firstly,analyze the background and significance of the subject research and the current situation of tunnel inspection vehicles at home and abroad,complete the system scheme design of the inspection vehicles used in this article,carry out the vehicle design for the tunnel inspection vehicle crack detection system,and carry out the vehicle and camera Selection of related hardware such as lens and lens.Secondly,analyze the error source of the crack detection system,establish the vehicle coordinate system;analyze the causes of image jitter,build the image motion model,complete the image homography matrix conversion and the camera distortion model analysis;the principle of tunnel image compensation and some classic electronics introduce learning compensation methods,and compare and analyze multiple feature point detection methods.Thirdly,a vehicle-mounted tunnel lining crack detection method based on image compensation is proposed.The unevenly illuminated image is processed by ambient light image enhancement to obtain an image with moderate brightness.Then the feature point detection based on the SIFT algorithm is performed.After the matching points are initially screened,use the adaptive RANSAC algorithm to accurately filter the matching points to complete the calculation of the global motion parameters;perform Kalman filtering on the global motion parameters to eliminate the irregular camera shake affected by vibration,and use the bicubic interpolation method to complete the tunnel image compensation;after compensation,the image is denoised by median filtering and bilateral filtering fusion,and the image segmentation is performed with Otsu dual threshold based on Canny edge detection;the image with noise after segmentation is denoised and morphologically processed based on regional shape features;cracks are processed Classification and feature extraction to obtain information such as crack length and width.Finally,build the laboratory truss of the crack detection platform,verify the camera and lens selection;introduce the calibration method in this article and complete the camera calibration work;collect data from the actual vehicle and process the tunnel image data,and finally the images before and after compensation.Do quality analysis and comparison of crack width.Experiments have proved that the vehicle-mounted tunnel lining crack detection system and method based on image compensation proposed in this paper can reduce the impact of crack detection system jitter,improve image quality,and have strong applicability to tunnel crack detection,which lays a foundation for the scientific research project supported by this article basis. |
PDF Full Text Request