Segmented self-contour reconstruction for rapid 3D modeling of oil and gas pipes

Abstract To solve the problem of low efficiency in the 3D modeling of pipes for dimensional defect detection in automotive engine oil and gas pipes, this paper proposes a method of segmented self-contour reconstruction for rapid 3D modeling of oil and gas pipes. Pipes formed by tube benders maintain a constant diameter and consist of two parts: straight pipe segments (SPSs) and arc pipe segments (APSs). The proposed method first constructs a single-view perspective projection model for SPS to solve its spatial axis and complete SPS reconstruction. Then, a single-view perspective projection model for APS is built to solve the center of its cross-sectional circle. By applying a homography transformation to convert the image plane into a front view plane, the APS axis is determined via circle fitting based on the SPS axis and the cross-sectional circle center in this plane, thereby completing APS reconstruction. Finally, the complete pipe is reconstructed by integrating the results from SPS and APS. In physical experiments, compared with the coupled-point-pair method, the proposed method achieves an 85.507% improvement in reconstruction accuracy while enhancing the reconstruction speed by 95.755%, demonstrating its effectiveness in achieving rapid 3D modeling of pipes.