Robust estimation of camera parameters using combinatorial optimization

The estimation of the parameters of the visual system is an in dispensable step for augmented reality or image guided applications where quantitative information should be derived from the images. Usually, the estimation process is called camera calibration and it is performed by observing a special calibration object from different directions. From these observations the image coordinates of the projected calibration marks are extracted and the mapping from the 3d world coordinates to the 2d image coordinates is calculated. To attain a well-suited mapping, the calibration images must suffice certain constraints in order to ensure that the underlying mathmatical algorithms are well-posed. Thus, the choice of the input images influences the estimation process and consequencely the quality of the derived information. In this paper we present a generic approach for camera calibration that is robust against ill-posed configurations. For this, we apply combinatorial optimization technique in order to determine the optimal subset of the pool of aquired images yielding the best calibration result with respect to the model fit error. Our approach is generic in the sense that it is independent of a certain calibration algorithm because it only makes use of a quality measure that acts as an objective function for the optimization.