Improving accuracy and confidence interval of camera parameters estimated with a planar pattern

Camera calibration is considered an important issue in computer vision. Methods based on different calibration patterns exist. Most common is the one which used a planar pattern because it uses a very simple setup and accurate results can be computed. Since points coordinates used to perform calibration process are corrupted with noise it is necessary to take into account its effects. Normally an interval in which camera parameter is included should be computed. In this paper an improvement of camera calibration process based on planar pattern is presented. It is based on using an appropriate weight to each data vector included in the estimation process. Each weight is computed depending on the quality of the input data. Moreover, a way to compute an interval in which camera parameters are including is presented.