A composite surface registration method for freeform surface evaluation based on ICP coarse registration and PSO fine registration

Abstract

The assessment of form error for freeform surfaces is crucial for ensuring the machining accuracy of optical components and performing potential subsequent error correction. Although the traditional iterative closest point (ICP) algorithm is widely used for surface registration due to its simplicity and broad applicability, it has limitations such as sensitivity to initial optimization positions, which seriously affect the accuracy of registration. For this reason, this paper proposes a composite surface registration method based on ICP for coarse registration and particle swarm optimization (PSO) for fine registration. First, the influence of six mounting errors on the form accuracy is discussed through error sensitivity analysis. Second, the characteristics of the ICP algorithm and the PSO algorithm in surface registration are comparatively analyzed in aspects of accuracy and efficiency. Finally, the composite registration method is successfully applied to the surface registration of the freeform surfaces. The results show that the composite registration method has achieved 88.4 % and 1.4 % decrease in registration error compared to the ICP algorithm when processing the freeform suction cup and silicon mirror, respectively. Additionally, the computational time is reduced by 40 % and 20 % compared to the PSO algorithm, respectively. Even when compared to advanced hybrid registration algorithms combining genetic algorithms (GA) and ICP algorithms, the proposed method in this paper still maintains advantages in terms of registration accuracy and efficiency.