3D Point Cloud Verification and Scatter Removal Using a Dual Camera/Projector Setup.

Area of Science:

• Robotics
• Computer Vision
• Optical Engineering

Background:

• Scattering in turbid water creates false 3D measurements, limiting underwater robotics.
• Uncertain object boundaries from scattering reduce the usefulness of 3D imaging data.
• Existing 3D imaging methods struggle with accuracy in challenging underwater conditions.

Purpose of the Study:

• To develop a robust method for precise 3D imaging in turbid water.
• To eliminate false 3D measurements caused by light scattering.
• To enhance the reliability of 3D data for underwater robotic applications.

Main Methods:

• Utilized triangulation-based 3D imaging principles.
• Analyzed camera/projector geometries and projected patterns for scattering discrimination.
• Developed an efficient O(1) algorithm for real-time false measurement identification.

Main Results:

• Achieved a 50-100x reduction in false positive rates compared to single-camera systems.
• Maintained a near-zero false negative rate across various turbidity levels.
• Demonstrated significant robustness improvements for subsea 3D imaging systems.

Conclusions:

• A combination of specific hardware, projected patterns, and algorithms enables reliable 3D imaging in turbid water.
• The developed method significantly enhances the trustworthiness of subsea 3D imaging systems.
• This advancement supports more complex and reliable underwater robotic operations.