A Rapid and Accurate Method for Point Cloud Reconstruction of NURBS Surface

The computational performance of the existing reverse engineering point cloud data processing methods in the aspects of mass point cloud multi-view fusion and NURBS free surface reconstruction needs to be improved. In order to overcome the shortcomings of point cloud multi-view fusion in computing efficiency, the principal curve of point cloud was extracted by data dimensionality reduction. By using geometric information such as curvature, torsion and spatial position of the principle curve, the transformation matrix between point clouds from different views is obtained. The method of micro-tangent plane in point neighborhood is used to calculate the point cloud vector. The NURBS surface is reconstructed by point cloud, and the reconstructed 3D surface model has high approximation accuracy and correct topology. The proposed algorithm has high efficiency and stability.