Flexible Camera Calibration Using a New Analytical Radial Undistortion Formula with Application to Mobile Robot Localization

-Most algor i thms in 3D computer vision rely on the pinhole camera model because of its simplicity, whereas virtually all imaging devices int roduce certain amoun t of nonlinear distort ion, where the radial dis tor t ion is the most severe part. Common approach to radial dis tor t ion is by the means of polynomial approximat ion, which introduces distortion-specific pa ramete rs into the camera model and requires es t imat ion of these dis tor t ion parameters . The task of es t imat ing radial dis tor t ion is to find a radial dis tort ion model tha t allows easy undis tor t ion as well as satisfactory accuracy. This paper presents a new radial dis tor t ion model with an easy analytical undis tor t ion formula, which also belongs to the polynomial approximat ion category. Experimental results are presented to show tha t with this radial distortion model, satisfactory accuracy is achieved. An application of the new radial dis tor t ion model is non-i terat ive yellow line al ignment with a cal ibrated camera on ODIS, a robot built in our CSOIS (See Fig. 1).