Three-dimensional reconstruction method for sparsely textured surfaces via freely moving line-structured-light-based...

Area of Science:

• Computer Vision
• Metrology
• Robotics

Background:

• Traditional stereo vision methods for 3D reconstruction are often static, limiting their measurement range and applicability.
• Sparsely textured surfaces pose challenges for conventional 3D reconstruction techniques due to difficulties in feature point extraction.

Purpose of the Study:

• To develop a novel 3D reconstruction method that overcomes the limitations of static stereo vision systems.
• To enable accurate measurement of sparsely textured surfaces using a freely moving line-structured-light binocular stereo vision setup.

Main Methods:

• Projecting line-structured light onto surfaces and capturing images with stereo cameras.
• Utilizing the Steger algorithm for precise light stripe extraction and binocular intersection for 3D coordinate calculation.
• Employing feature point-based relative pose estimation and reprojection error minimization for accurate camera pose determination.

Main Results:

• Successfully reconstructed the 3D shapes of two metallic surfaces with varying geometric configurations.
• Demonstrated generalizability across different surface types, indicating robustness of the proposed method.
• Achieved high accuracy in 3D reconstruction, with average vertical distances not exceeding 1.52 mm and standard deviations below 0.97 mm.

Conclusions:

• The proposed freely moving line-structured-light binocular stereo vision method significantly enhances the measurement range and applicability of 3D reconstruction.
• The technique is effective for reconstructing sparsely textured and metallic surfaces, proving its versatility.
• The method offers a robust and accurate solution for dynamic 3D shape measurement in computer vision applications.