3D Reconstruction of Nonuniform Rational B-Spline Surface Based on Line-Structured Light

Abstract

Structured light is considered one of the fundamental and reliable techniques for 3D surface reconstruction. In particular, the line-structured light reconstruction method has been widely adopted due to its high precision, robust stability, and simple design. However, using line-structured light to reconstruct complex surfaces, particularly those represented by nonuniform rational B-Spline (NURBS), presents significant challenges. In view of this, this paper focuses on the 3D reconstruction of NURBS surfaces and does the following work: First, the R-channel stripe center extraction algorithm is presented because the laser stripes used in this process often suffer from severe deformation, which can lead to difficulties in center extraction and discontinuity issues. Second, the 3D reconstruction of the NURBS surface is carried out. The average absolute error of the reconstructed surface in the x-direction is 0.1244mm, and the average relative error is 0.11%, reflecting the accuracy of the reconstruction algorithm in this paper. Finally, an improved Iterative Closest Point (ICP) algorithm is presented to overcome the challenge of point cloud registration caused by the similarity in shape between the peaks and valleys of NURBS surfaces. Experimental results show that it significantly improves the registration quality of NURBS surfaces.