Automatic multi-stereo-vision reconstruction method of complicated tubes for industrial assembly

Abstract

Purpose Complicated tubes extensively exist in the industrial equipment. The manufacturing precision of the tubes is difficult to be ensured with bending machine. Therefore, the tubes’ 3D geometric error should be fixed according to measurement results. However, there are no convenient methods to accomplish the measurement accurately and effectively. Thus, this paper aims to propose a new tube measurement method to achieve tube's automatic measurement. The accurate measurement results can be used to fix the geometric error of the tube to achieve stress-free assembly. Design/methodology/approach Tubes’ shape can be determined by control points. First, the point clouds of the centre line by multi-stereo-vision technology are reconstructed. Then, the point clouds to the spine of the tube are thinned by moving least-squares and segmented into lines and arcs. Finally, the control points are calculated and the model is reconstructed. The authors can get the tube’s geometric dimensions from the model. Findings The experiment results indicate that the multi-stereo-vision technology can solve the occlusion problem and measure the complicated tubes efficiently and accurately. Originality/value The paper proposed a tube measurement method. The repeatability measuring precision was 0.12 mm, and the absolute measuring precision was within 0.78 mm. The tube spectra assessed in this paper are in the range of angles between two adjacent line segments of 3-177° and the shortest length of the line segment is greater than 5 mm, confirming that the proposed algorithm can measure various complicated tubes effectively and accurately.