Fast and practical method for underwater stereo vision calibration based on ray-tracing.

Area of Science:

• Computer Vision
• Robotics
• Optics

Background:

• Standard camera models struggle with underwater light refraction.
• Accurate calibration is crucial for underwater scene reconstruction.

Purpose of the Study:

• To propose an accurate and convenient calibration method for large underwater scenes.
• To address the limitations of perspective camera models in underwater environments.

Main Methods:

• Developed an underwater refractive camera model.
• Utilized refraction coplanarity constraints for initial parameter estimation.
• Employed co-point constraints for nonlinear optimization, simplifying existing approaches.

Main Results:

• Achieved reconstruction accuracy below 0.02 mm in a 200m x 200m field of view.
• Demonstrated effectiveness in large-scale underwater environments.
• Showcased equal effectiveness with sparse calibration data.

Conclusions:

• The proposed underwater refractive camera model enables accurate and convenient calibration.
• The method offers significant improvements in reconstruction accuracy for underwater applications.
• This approach simplifies the calibration optimization process.