This work deals with peculiarities of representation method for visual scene transforms in task of a mobile robot video scene analysis. This method permits to avoid necessity to know parameters and localization of a video sensor. Coefficients of Lie transformations for the direct and inverse mapping “scene – cadre” (or 'scene – scene', during the robot motion) are got as result of a solving of calibration equations. An example of visual data interpretation is given for visual system of mobile robot “Trikol”.