ORB-SLAM3-Enhanced 3-D Reconstruction via Underwater Teleoperator With Global Prescribed Performance

Underwater three-dimensional (3D) reconstruction is of great significance for underwater structure detection. However, ocean current disturbance and color distortion pose significant challenges in achieving reliable 3D reconstruction of underwater scenes. This paper investigates 3D reconstruction based on movable stereo camera via an underwater teleoperator. Specifically, a vision system based on ORB-SLAM3 is constructed to calculate the depth maps and perform dense 3D reconstruction through real-time video frames. An improved Multi-Scale Retinex with Color Restoration-Color Correction (MSRCR-CC) algorithm is proposed to address color distortion in underwater image enhancement. A redundant point removal algorithm is designed for the point cloud mapping process, effectively reducing the system's computational burden. Furthermore, to ensure stable scanning and obtain a high-quality original video stream in disturbed underwater environments, a disturbance observer-based prescribed performance control scheme is proposed for the underwater teleoperation system. The proposed control strategy guarantees that the master- slave synchronization error is bounded within predefined boundaries, which can enhance the fast fourier transform response value of the original image, and thus improve the 3D reconstruction quality. A global prescribed performance function is applied, which removes the initial value restriction on synchronization errors. Finally, simulation and experimental results are provided to illustrate the effectiveness of the proposed methods.