A Robust Method for Camera Calibration in Noisy Settings Based on Genetic Programming

In this article, we introduce a novel camera cali-bration method using genetic programming, to calibrate cam-eras in noisy environments. Traditional calibration methods, such as those of Tsai and Zhang, extensively use the pinhole camera model, and are less accurate in the presence of noise. In this work, high precision calibration is achieved using pseudo linear genetic programming. Instead of the pinhole camera model, pseudo linear genetic programming generates mathematical functions which allow for far greater precision in the calibration process than classical methods, regardless of the environmental conditions. The method has several challenges such as identification of suitable calibration functions and creation of an extensive training database. However, the method provides advantages in terms of better results quality and practicality, as it eliminates the necessity of the intrinsic camera parameters. The results illustrate that this methodology is far superior in comparison to the current state-of-the-art technique, Zhang's widely used method, with a 20x improvement in calibration accuracy.