Three-dimensional measurement of small inner surface profiles using feature-based 3-D panoramic registration

Area of Science:

• Metrology
• Machine Vision
• Optical Engineering

Background:

• Three-dimensional (3-D) optical measurement offers advantages like noncontact, high accuracy, and automation for manufacturing quality control.
• Current optical methods are limited to external surfaces, leaving a gap in inspecting complex internal features of machined parts.

Purpose of the Study:

• To propose and validate a novel machine vision-based approach for 3-D profile measurement of tiny, complex internal surfaces.
• To enable accurate topographic characterization of internal features, such as internally threaded holes.

Main Methods:

• Utilized a side-view rigid scope to capture images of internal surfaces from known camera positions.
• Generated 3-D point clouds using multiview stereo vision combined with linear and rotational motion of the test piece.
• Developed an automated feature-based 3-D registration algorithm to combine multiple point clouds into a complete surface reconstruction.

Main Results:

• Successfully generated a complete 3-D reconstruction of tiny complex internal surfaces.
• The proposed method demonstrated feasibility when compared against X-ray computed tomography for validation.

Conclusions:

• The developed machine vision approach is effective for 3-D optical measurement of intricate internal geometries.
• This method holds potential for future robotic-driven industrial inspection, enhancing quality control capabilities.