Accurate calibration of a multi-camera system based on flat refractive geometry.

Area of Science:

• Computer Vision
• Optical Engineering
• Metrology

Background:

• Multi-camera systems are essential in various fields, but accurate camera calibration remains a significant challenge.
• Common setups involve cameras on opposing sides of an object with overlapping fields of view.
• Calibration errors can arise from using transparent calibration targets due to light refraction.

Purpose of the Study:

• To present a novel multi-camera calibration method.
• To address and eliminate calibration errors caused by light refraction through a transparent glass calibration board (TGCB).
• To accurately determine intrinsic and extrinsic camera parameters for multi-camera systems.

Main Methods:

• Utilizing flat refractive geometry theory to correct for refraction-induced errors.
• Integrating the camera projection model with flat refractive geometry.
• Employing the bundle adjustment method to minimize reprojection error and optimize calibration results.
• Simulating performance with varying levels of Gaussian noise.

Main Results:

• Simulations demonstrated high accuracy, with rotation angle error < 5.6e-3 degrees and translation error < 4.6e-3 mm.
• Real-world experiments with a four-camera system yielded a mean reprojection error of 4.3411e-05 pixels and a standard deviation of 0.4553 pixels.
• The proposed method proved accurate and feasible under various noise conditions.

Conclusions:

• The novel method effectively corrects for refractive errors in multi-camera calibration.
• The approach achieves high precision in determining camera intrinsic and extrinsic parameters.
• Both simulation and experimental results validate the accuracy and feasibility of the proposed calibration technique.