Depth Sensing by Near-Infrared Light Absorption in Water

Area of Science:

• Optics
• Computer Vision
• Fluid Dynamics

Background:

• Accurate 3D depth recovery is crucial for underwater robotics and imaging.
• Existing methods struggle with varying light conditions and object surface properties in water.

Purpose of the Study:

• To introduce a novel bispectral depth recovery method for underwater objects.
• To demonstrate the robustness and accuracy of the proposed technique across different scenarios.

Main Methods:

• Utilized the Beer-Lambert model and light absorption differences at two near-infrared wavelengths.
• Developed algorithms to correct for practical implementation nonidealities.
• Constructed a coaxial bispectral depth imaging system with off-the-shelf hardware.

Main Results:

• Achieved accurate depth recovery irrespective of surface texture and reflectance.
• Successfully demonstrated shape recovery of complex and dynamic underwater objects.
• A trispectral variant showed improved robustness for challenging surface reflectances.

Conclusions:

• The proposed "shape from water" method offers a viable and robust solution for underwater 3D depth recovery.
• The system is practical, cost-effective, and validated by experimental results.