Robot vision modelling-camera modelling and camera calibration

Abstract

Camera modelling, especially parameter calibration, is a crucial problem for further progress in robot vision applications. Simply stated, the problem of calibration is how to evaluate the position and orientation of the camera easily, rapidly, accurately, and reliably. After the importance of camera modelling is explained, this paper will review the state of the art of calibration techniques. Categorized calibration techniques are described and characterized in detail. A common set of symbols is adopted and applied to the various techniques, and complementary derivations are partly made to clarify the peculiarities and problems of each technique. Detailed considerations on the preservation of the orthogonality constraints for the rotation transformation matrix and on the instability of the calibrated intrinsic parameter values using real experimental set-ups are expected to pave the way for future research on robot vision applications.