Underwater Target Detection and 3D Reconstruction System Based on Binocular Vision

Area of Science:

• Robotics
• Computer Vision
• Marine Technology

Background:

• Underwater target detection and 3D reconstruction present significant challenges.
• Existing methods struggle with the complexities of marine environments and 3D data acquisition.

Purpose of the Study:

• To develop a robust binocular vision system for underwater target detection.
• To improve the accuracy and effectiveness of 3D reconstruction for underwater targets.

Main Methods:

• Utilized a binocular vision system with two optical sensors.
• Applied image denoising, color restoration, and super-pixel segmentation for target extraction.
• Improved semi-global stereo matching with constrained matching and least squares fitting for disparity map optimization.
• Employed triangulation principles for 3D data calculation and MeshLab for visualization.

Main Results:

• The system demonstrated higher measurement accuracy for specific underwater targets.
• Achieved a superior 3D reconstruction effect within a suitable operational distance.
• Successfully extracted and reconstructed target structures and color information.

Conclusions:

• The proposed binocular vision system effectively addresses underwater target detection and 3D reconstruction challenges.
• The optimized stereo matching and 3D calculation methods enhance accuracy and visual fidelity.
• The system is suitable for applications requiring precise underwater target measurement and modeling.