High accuracy camera calibration for 3-D computer vision

A method for calibrating intrinsic TV camera parameters such as effective focal length, one-pixel length on the image plane, the image distortion center, and the distortion coefficient is presented. An ordinary two-dimensional model fitting is decomposed into two one-dimensional model fittings on the column and row of a frame buffer near the image distortion center by utilizing the point symmetry characteristic of picture distortion. Some parameters are also eliminated by setting up a calibration chart. Thus, the number of parameters to be fitted is drastically decreased, enabling fast and accurate calibration. Calibration accuracy is related to calibration conditions by computer simulation and is confirmed by experimentation.<
>