Design Of Crack Detection System For Inner Wall Of Small-Diameter Metal Pipes

Cracks on the inner wall of metal pipes are often caused by stress or corrosion.If not monitored,the cracks will grow and cause the pipe to rupture,causing irreparable losses.To solve such problems,pipeline cracks need to be regularly inspected to monitor whether the cracks are growing,whether they need to be repaired or replaced,so fast,accurate,and stable requirements are placed on the inspection system.In this thesis,a crack detection system for the inner wall of small-diameter metal pipes is designed based on panoramic images.The combination of software and hardware realizes the accurate extraction of crack feature information,which meets the requirements in the final system performance verification experiment.The thesis first discusses various methods of crack detection.After comparing and summarizing the applicability of the existing technology,a fisheye lens is used to obtain a panoramic image,and an image processing algorithm is used to obtain a complete inner wall image.Check and theoretically analyze the feasibility of the system.Finally,through the selection of key devices and the design of hardware parameters,the construction of the system device was successfully completed.In view of the poor quality that may exist in the actual image,the concept of image preprocessing is proposed.Through the comparison of effects,the combination of adaptive median filtering and histogram equalization was determined to enhance the image contrast,and Zhang Zhengyou's calibration of the camera to correct image distortion and improve image quality.Next is the determination of the relevant image processing algorithms,including the unfolding algorithm of the ring image,the stitching algorithm of the reconstructed image,and the crack detection algorithm.Based on the comparison of MATLAB experimental simulation results,the optimized algorithm uses the optimized version of the concentric circle expansion method with faster operation speed and better expansion quality;the stitching algorithm uses a combination of Harris corner matching algorithm based on ROI and the fade-in and fade-out algorithm.To achieve the distortion-free stitching of the reconstructed image;the crack detection algorithm uses an improved,ROI-based Canny edge detection algorithm to achieve accurate extraction of crack edges.Finally,the performance of the crack detection system is quantitatively analyzed,and related verification experiments are designed.Through analysis of the mean absolute error,root mean square error,standard deviation,and error based on statistical theory of the data,the crack detection system of the inner wall of the inner diameter metal pipe based on the smalldiameter has a measurement accuracy of 1mm and a high stability in this thesis.