Research On Multi-vision-based Digitized Measurement Method For Pipelines

Pipelines are widely applied to the complex electromechanical products in industrialfields of aviation, aerospace and automotive, and their reliable assembly directly affectsproduct’s quality. Measurement is one of the direct and effective ways to guaranteepipeline’s processing quality and non-stress reliable assembly. With the development ofcomplex electromechanical products in the direction of lightweight, miniaturization andstructural integration, the space packing density of products becomes much higher, and theshape of pipelines becomes more complex, which makes measurement for pipelines moredifficult. The traditional measurement methods for pipelines can not meet the currentengineering requirements. To solve the fast and precise non-contact measurement problemfor pipelines, the basic principles and realization techniques of multi-vision-based dititizedmeasurement method for pipelines are deeply studied. A prototype system is developed andits application verification is carried out.The research’s main results are summarized as follows:(1) The framework and camera calibration method of the multi-vision-based digitizedmeasurement system are studied. A multi-vision-based dititized measurement method andsystem for pipelines are present. The measurement method’s technical system and themeasurement and reconstruction process of pipeline’s pipe and joint are respectivelyformulated. The principle of camera imaging model and multi-vision system is analyzed.Through the difference analysis and experiment of two calibration plates, a multi-cameracalibration method is provided based on grouping.(2) A measurement and reconstruction method for pipeline’s pipes based on centerlinematching and registration is present, through analyzing pipeline’s features. The perspectiveinvariance of NURBS curve is analyzed. The extraction method of pipeline’s centerlinesand feature points is provided. NURBS curves are selected as the description approach andmatching primitives, and the matching and reconstruction method of pipeline’s centerline isprovided. For branched pipelines in large size, the registration method of pipeline’scenterlines is provided to realize precise and fast measurement and reconstruction.(3) A measurement and reconstruction method for pipeline’s joints based on shape features is present, through analyzing the problem of joint measurement and reconstruction.Shape features are selected as the matching features.2D projected images are acquiredfrom numerous views of virtual cameras and layered in the pyramidal structure to composeshape feature images based on the CAD model of joint. The joint’s edges images arematched with its shape feature images to determine its precise pose. The calculationformula of the joint’s pose parameters is derived to realize precise and fast measurementand reconstruction.(4) The error analysis and modeling method for pipeline measurement system isstudied. The measurement process is analyzed to establish the error transmission model.The calibration accuracy effects of calibration point’s number, size, calibration image’snumber and noises are analyzed, and a better calibration scheme is provided. Thetheoretical calculation formulas of matching and reconstruction errors are derived. Ameasurement error model based on calibration parameters perturbation is present andestablished, and improving methods of measurement accuracy are provided. Simulationexperiments are carried out to analyze the factors which affect the errors of centerlineextraction, image parallel rectification and reconstruction. The results of measurementexperiment prove the present error model’s availability.(5) On the basis of the research results and enterprise needs, a pipeline multi-camerameasurement system is developed, and the system is applied to pipelines measurement andreconstruction to verify the present measurement methods for pipeline’s pipes and joints.The system improves pipeline measurement efficiency and meets accuracy requirement,which means the present measurement method and system have the engineering applicationvalue.