A precision reconstruction method based on refined and adjusted multilevel T-spline for complex surface manufacturing

Abstract

Precision reconstruction is a critical technology in complex part manufacturing and measurement. The rapid development of data acquisition systems, represented by the 3D laser scanner and Coordinate Measuring Machine (CMM), requires simulation and reconstruction capabilities for large-scale, dense, and non-uniform point clouds. Moreover, advancements in T-spline theory enable a significant reduction in control points while maintaining high accuracy. This paper proposes a Refined and Adjusted Multilevel T-spline method based on the point-to-surface shortest Distance (D-RAMT) to reconstruct surfaces from measured point clouds. Reconstruction performance is significantly improved via alternating cycles of multilevel adjustment and refinement. Point-to-surface shortest distances are computed as reconstruction errors using a proposed algorithm combining Random points, Set-inherited search, and Parallel computation (RSP). Additionally, an Improved B-spline Approximation (IBA) method for solving control points and an adaptive T-mesh local refinement strategy are developed. Machined workpieces with complex surfaces are utilized to verify the proposed method. Results show that the D-RAMT has better reconstruction capability with high accuracy, acceptable convergence times, and fewer control points (37–62 %) than other methods.