Research On Oil And Gas Pipeline Weld Detection And Identification System Based On Machine Vision

With the continuous advancement of oil and gas pipeline construction,pipeline safety has become an increasingly important focus.Pipeline welds are the key to long-term and safe transportation of pipelines.Therefore,to ensure the quality and safety of pipeline welding,project personnel need to inspect the oil and gas pipeline weldments.At present,the common detection methods are mainly manual detection.With the increase of pipeline detection points,manual detection exposes problems such as low detection efficiency and easy detection of subjective judgment results.This thesis introduces machine vision related technologies to the inspection and identification of oil and gas pipeline welds,which has some guiding significance for the inspection and identification of oil and gas pipeline welds.1)Based on the research of machine vision image preprocessing,feature extraction and other theories,this paper analyzes the problems of oil and gas pipeline welds based on field data,such as poor image contrast,blurred surface texture,and serious noise pollution.The characteristics of the weld image,such as feature concentration,obvious hierarchy,and uneven weld distribution in the image.The intelligent positioning model of the weld based on the characteristic energy distribution and an improved non-local mean filtering algorithm are studied and proposed to realize the weld region of the oil and gas pipeline.Realize accurate positioning of the weld area of the oil and gas pipeline,regionalize noise reduction of the weld image,and achieve the effect of intelligent noise reduction.2)Based on in-depth study of curve fitting related technologies and theories,detailed analysis of the surface texture characteristics of pipeline weld image,research and put forward an algorithm for surface morphology detection of oil and gas pipeline welds based on curve fitting.The discrete point fitting and the calculation of the similarity coefficient have realized the detection of the surface morphology of the oil and gas pipeline welds.3)This paper studies image segmentation technology in depth,and combines the characteristics of oil and gas pipeline weld defects,such as blurred edges and different shapes,to design and implement a set of extraction and recognition methods for oil and gas pipeline weld defects.This method is based on pre-processed image data,using binarization method,snake model and other methods to theoretically detect,segment and accurately locate the defect edges of oil and gas pipeline welds.It proposes to extract defect perimeter,defect area,circularity,and eccentricity.The six major parameters,area radius,and surface similarity coefficient of the weld are used to express the defect features,and then the KNN algorithm is used to classify and identify the defect parameters,thereby realizing the detection and identification of weld defects in oil and gas pipelines.4)Design and implementation of oil and gas pipeline weld detection and identification system.The object-oriented development language is used to design and implement a machine vision-based oil and gas pipeline weld detection and recognition system,which provides a certain guidance for the oil and gas pipeline weld detection and recognition.