Research On Pipeline Crack Detection Based On Video Image

As an important transportation resource facility,oil and gas pipelines provide great convenience for sustainable energy development and greatly save energy transportation costs.However,during the long-term use of pipelines,various types of defects,such as deformation,cracks,and corrosion,will occur.Regular inspections of pipeline defects and pipeline maintenance are very important.Video image pipeline defect detection system is a commonly used detection system at home and abroad.It needs to be manually assisted in the actual detection process.Therefore,a large amount of artificial uncertainties are added to the detection process,and its reliability is poor,resulting in low efficiency of defect image detection.,error and so on.Aiming at the shortcomings of the traditional video pipeline defect detection system,this paper develops a pipeline inner wall video image acquisition robot and CCD video image detection system.The video image information of the inner wall of the pipeline is collected in real-time by the CCD camera detection device mounted on the robot.The wireless transmission module transmits the data to the host computer system.The host computer system performs a series of processing on the pipeline image,and finally realizes feature extraction and automatic detection of pipeline defect images.In this paper,a video image detection pipeline robot which adapts to small and medium pipe diameters is developed.The pipeline robot's drive mode,power supply,and video image detection and acquisition scheme are analyzed and designed.According to the characteristics of the actual pipeline image acquisition process and the real-time requirements of the detection system,the LED lighting source design,CCD wide-angle digital camera and video controller chip selection were completed.Pipeline video image acquisition,compression,and data transfer interface protocols were set up.The development of the human-machine interaction interface of the host computer was completed,real-time detection,transmission and display of video image information on the inner wall of the pipeline was realized,and the subsequent image processing and pipeline were completed.Defect inspection laid the foundation.Aiming at the special inner environment of the pipeline and the complex background of video images,the noise characteristics of the pipeline defect images are analyzed,and a method for preprocessing the pipeline defect images is proposed.The pipeline defects are improved by using image denoising and image enhancement methods.The quality of the image,and through the binarization process of dynamically selecting the threshold between the maximum variance between classes,the defect feature information is highlighted from the background image of the pipeline.The feasibility of pipeline defect pretreatment was verified by experimental results.On the basis of pipeline defect image preprocessing,the boundary of the defect image is determined,and the traditional Canny edge detection operator is improved to avoid the occurrence of false edges and preserve the useful edge information.This paper presents a new edge segmentation based mathematical morphology segmentation based on edge detection(MSED),and describes the implementation principle of the algorithm in detail.For the two typical pipeline defects of pipeline interface cracks and cracks,the morphological open top hat operator,closed bottom hat operator and MSED operator are used to process the defect image data.Finally,the experimental results show that based on mathematical morphology The open top hat algorithm is suitable for the detection of seam flaws in pipelines and has good robustness;the morphology based on edge detection number can effectively detect the cracks in pipelines.For the description of pipeline defect shape,feature parameters such as defect area,long axis,short axis and eccentricity are selected to complete the derivation calculation.The quantitative calculation results of the defect characteristics are evaluated by analyzing the completeness,correctness,and defect detection quality of the actual defect features.