Self-Calibrated In-Process Photogrammetry for Large Raw Part Measurement and Alignment before Machining

Area of Science:

• * Industrial Metrology
• * Computer Vision
• * Photogrammetry

Background:

• * Photogrammetry is increasingly used for 3D metrology in industry, relying on optical targets and pre-calibrated cameras.
• * Traditional post-process bundle adjustment is computationally intensive, leading to lengthy, iterative measurement workflows.
• * Existing methods require precise off-process camera calibration, limiting flexibility and potentially introducing uncertainty.

Purpose of the Study:

• * To develop an efficient in-process procedure for bundle adjustment in portable photogrammetry.
• * To enable quasi real-time 3D scene computation on standard hardware.
• * To integrate self-calibration for camera and lens distortion, enhancing precision with low-cost cameras.

Main Methods:

• * Implemented an in-process bundle adjustment algorithm for photogrammetry.
• * Integrated a self-calibration method for camera and lens distortion parameters.
• * Utilized scale bars within the scene for measurement traceability, eliminating reliance on external calibration.

Main Results:

• * Demonstrated in-process bundle computing capability on a consumer-grade PC, achieving computation times under 1-2 seconds per image.
• * Achieved a relative precision of 1/10,000 (0.1 mm error in 1 m) with an image plane RMS error below 0.2 pixels.
• * Validated the method in pilot and industrial raw part measurement scenarios, matching performance of traditional methods.

Conclusions:

• * The developed self-calibrated in-process photogrammetry offers a fast, precise, and user-friendly solution for industrial measurements.
• * This approach reduces reliance on specialized equipment and complex calibration procedures.
• * Enables near real-time 3D data acquisition and processing for applications like raw part inspection.