Study Of Pipe Defect Recognition And3D Reconstruction Methods Based On Circle Structured Light Vision

Pipe is the most secure and efficient mode in transport of gases andliquids. Pipe inner surface defects detection has an important significance toextend pipe life, reduce waste and prevent accidents. Traditional pipenon-destructive defect testing methods have low detection efficiency, and theyare not visible. Visual detection technology which based on circle structuredlight can finish accurate3D measurement on the pipe inner surface. Afterfinishing pipe defect detection by processing3D data points, the3Dreconstruction technology is used to reconstruct the pipe inner surface, inorder to realize the three-dimensional morphology visualization of the pipeinner surface. It has a very important practical significance and industrialvalue.This thesis centers around the National Natural Science Foundationproject-Pipe inner surface3D morphology visual detection theory andtechnology studies, based on pipe inner surface3D point cloud data whichacquired by circle structured light vision inspection system, to study pipedefect recognition and pipe inner surface3D reconstruction.First, the3D point cloud data reduction methods were studied. K-Nearest Neighbor3D data reduction algorithm was used to reduce pipe inner surfacepoint cloud data. Based on the characteristics of pipe inner surface circlestructured light vision3D data, a method was proposed to recognize pipe innersurface convex defect and pit defect by comparing the normal vector anglebetween adjacent points which were on the same circumference. In order torecognize hole, a distance constraint was proposed, that is, on adjacent twocircumferences, if the distances between corresponding adjacent two pointswere all smaller than a certain threshold, then we viewed this place as anon-hole, or else we viewed it as a hole. Then, the threshold in the holeboundary was reduced to recognize the hole boundary. Based on Delaunaytriangulation principle, research on the triangulation method of pipe innersurface circle structured light vision3D point cloud data which with hole andwithout hole, and to realize3D reconstruction last.Based on VC++6.0software platform, pipe inner surface defectsimulation data, hole simulation data and real3D data was used to verify thecorrectness of pipe defect identification method. Finally, OpenGL was usedfor surface reconstruction to achieve three-dimensional visualization.