The problem of calibrating cameras is extremely important incomputer vision. Existing work is based on the use of a calibrationpattern whose 3D model is known a priori. The authors present a completemethod for calibrating a camera, which requires only point matches fromimage sequences. The authors show, using experiments with noisy data,that it is possible to calibrate a camera just by pointing it at theenvironment, selecting points of interests, and tracking them in theimage while moving the camera with an unknown motion. The cameracalibration is computed in two steps. In the first step the epipolartransformation is found via the estimation of the fundamental matrix.The second step of the computation uses the so-called Kruppa equations,which link the epipolar transformation to the intrinsic parameters.These equations are integrated in an iterative filtering scheme
展开▼
