Global calibration method for non-overlapping cameras based on rigidly connected stereo targets.

Area of Science:

• Computer Vision
• Metrology
• Optical Engineering

Background:

• Single cameras have limited field of view (FOV) for global calibration.
• Traditional multi-camera calibration in large FOV environments suffers from overlapping FOV and complex targets.
• Existing methods like laser projection and plane mirrors are sensitive to environmental conditions.

Purpose of the Study:

• To propose a non-common FOV binocular calibration method for large FOV environments.
• To overcome limitations of overlapping FOV and complex target requirements in existing calibration techniques.
• To provide a simpler and more robust calibration solution for vision systems.

Main Methods:

• A rigidly connected stereo target composed of two checkerboard plane targets is utilized.
• Vision sensors capture images of sub-targets, and the target is moved multiple times.
• Spatial constraints of the rigidly connected target are used to derive the transformation relationship between sensors.

Main Results:

• The proposed method effectively calibrates non-common FOV systems.
• Experimental results show a Root Mean Square (RMS) error of 0.16 mm at a 13 mm distance.
• The method demonstrates effectiveness, simplicity, and independence from complex targets or large measuring instruments.

Conclusions:

• The developed binocular calibration method is effective for large FOV measurements in non-common FOV scenarios.
• It offers a simpler, more robust alternative to existing calibration techniques.
• The method successfully addresses the challenges of measurement in non-overlapping FOV environments.