A hierarchical registration method for optical surface measurement based on a meta-heuristic algorithm

Abstract Optical surface measurement and characterization precision are significantly influenced by the performance of surface registration methods. Most existing optical surface registration methods suffer from poor efficiency or lower precision. To address these challenges, this investigation proposes a hierarchical registration method based on a meta-heuristic algorithm, primarily targeting freeform and grid surfaces (relatively simple non-rotationally symmetric surfaces) and rotationally symmetric surfaces in optical applications. First, to enhance efficiency and avoid unnecessary registration, the importance of six degrees of freedom (DOFs) is evaluated and classified into four categories. The registration process then follows the importance level of the DOFs, prioritizing those of higher significance. The deviation value is obtained by considering the last registration result as a constraint condition and employing a meta-heuristic algorithm to minimize the fit function. Simulation and experimental results demonstrate that the proposed method achieves lower average, minimum, and standard deviation values compared to other registration methods. Notably, when the population size and the number of iterations are reduced, the standard deviation decreases by over 99% compared to conventional methods. The proposed method exhibits significant advantages in both precision and efficiency, and its practicality has been validated through experiments on freeform, aspheric, and sinusoidal grid surfaces.