Extrinsic Calibration of Camera Networks Using a Sphere

Area of Science:

• Computer Vision
• Robotics
• Geometric Calibration

Background:

• Accurate camera network calibration is crucial for multi-camera systems.
• Existing methods often rely on complex feature detection or epipolar geometry.
• Sphere-based calibration offers potential for simplification and robustness.

Purpose of the Study:

• To propose a novel and simplified extrinsic calibration method for camera networks.
• To enhance accuracy and robustness in camera parameter estimation.
• To reduce computational complexity compared to existing techniques.

Main Methods:

• A novel method for estimating 3D sphere center positions in local camera coordinates.
• Utilizing orthogonal Procrustes analysis for initial pairwise extrinsic parameter estimation.
• Employing an optimization routine for joint refinement of all camera extrinsic parameters.

Main Results:

• The proposed method achieves more accurate extrinsic parameter estimation than other sphere-based approaches.
• It requires simpler image processing (area and center of mass) compared to outline analysis.
• Demonstrates robustness, flexibility, and ease of use in experimental results.

Conclusions:

• The novel sphere-based extrinsic calibration method offers significant advantages in accuracy and simplicity.
• It provides a viable alternative to point-based feature and epipolar geometry methods.
• The approach is well-suited for practical applications requiring efficient and reliable camera network calibration.