Abstract
Large-aperture optical surfaces are difficult to measure at full aperture because of the limited aperture of interferometers. Although subaperture stitching can reconstruct the full-aperture surface, its accuracy is often degraded by subaperture positioning errors. To address this issue, we propose TRR-FPC (full phase calibration), a subaperture stitching method based on the trust region reflective (TRR) algorithm. A six-degree-of-freedom (6-DOF) rigid-body transformation model is established to characterize the relative poses among subapertures, and an objective function based on overlap-region residuals is optimized to estimate and compensate for pose errors. By jointly optimizing all subapertures in a unified coordinate system, TRR-FPC reduces error accumulation during stitching. Numerical simulations and optical flat stitching experiments were conducted to evaluate the proposed method. Experimental results show that, compared with least-squares (LSQ) and iterative closest point (ICP) methods, TRR-FPC reduces the peak-to-valley (PV) and root-mean-square (RMS) values of the difference phase by 13–20%, while better preserving low-order Zernike structure. These results indicate that TRR-FPC effectively suppresses pose-induced stitching errors and improves full-aperture reconstruction accuracy.