Profile reconstruction method adopting parameterized re-projection errors of laser lines generated from bi-cuboid...

Area of Science:

• Computer Vision
• Metrology
• 3D Reconstruction

Background:

• Accurate 3D profile recovery is crucial for various industrial applications.
• Existing methods often struggle with large fields of view and require complex calibration.
• Flexible and robust reconstruction techniques are needed for real-world scenarios.

Purpose of the Study:

• To develop a flexible and accurate method for 3D profile recovery using laser plane re-projection errors.
• To address challenges associated with large camera fields of view and parameter calibration.
• To validate the proposed method's performance and investigate influencing factors.

Main Methods:

• Utilized bi-cuboid references for a large camera field of view.
• Employed RQ decomposition for initial camera parameter estimation.
• Developed a balance model for global parameter matrix refinement.
• Solved the flexible laser plane using Plücker matrices and homographies.
• Implemented an optimization method based on re-projection errors for parameter refinement.

Main Results:

• Achieved average reconstruction errors of approximately 1.14 mm across different experimental conditions.
• Demonstrated the validity and accuracy of both the initial and optimized reconstruction methods.
• Investigated the impact of measurement, reference, and test distances on reconstruction accuracy.

Conclusions:

• The proposed flexible reconstruction method is effective for accurate 3D profile recovery.
• The optimization approach significantly improves laser plane and parameter matrix accuracy.
• The method shows promising application prospects in various fields requiring precise 3D measurements.