High-accuracy underwater binocular line-laser 3D measurement via multi-medium refractive calibration

Underwater optical three-dimensional (3D) measurement is fundamentally challenged by multi-medium refraction at air-glass-water interfaces, which invalidates the conventional central projection model and significantly degrades reconstruction accuracy. To address this issue, we propose an underwater binocular line-laser 3D measurement method that combines explicit refractive imaging modeling, refractive parameter calibration, and subpixel laser-stripe reconstruction. First, an underwater binocular line-laser imaging model is established for the air-glass-water optical path, in which the refractive ray paths are described using Snell's law, and the key parameters include the relative pose of the two cameras, the interface normal, the distance from the camera center to the glass interface, the glass thickness, and the refractive indices of the involved media. Second, a multi-objective optimization framework is developed to calibrate the refractive parameters by jointly minimizing the shortest distance between refracted binocular rays and enforcing geometric constraints of calibration-board corners, including equal edge length, coplanarity, and orthogonality. Third, for underwater line-laser reconstruction, a subpixel laser-stripe processing strategy is introduced by combining Laplacian-of-Gaussian filtering, zero-crossing edge localization, inscribed-circle fitting of stripe boundaries, B-spline smoothing, and 3D reconstruction under the refractive epipolar constraint. Simulation and experimental results demonstrate that the proposed calibration method substantially improves robustness against image noise. In real experiments, the developed underwater binocular line-laser scanner achieves a measurement accuracy better than 0.14 mm within a measurement range of 400 mm, showing its effectiveness for high-accuracy underwater 3D measurement.