The minimum distance measure of stereovision

The authors propose a method of online self-calibration of the general stereovision camera model. The underlying assumptions are that a set of homologous image pairs have been found in the stereovision matching process and that only the shifts in extrinsic camera parameters occur. The principle of the self-calibration method is to minimize the variance of a cost function which is defined as the mean square of the minimum distance measure. The stereovision cameras were simulated on the mainframe. The first study was to ascertain if there is a direct relation between the positional error and the MDM. In the second experiment, the self-calibration algorithm was implemented and tested for convergency. The results are discussed