The model of the ideal video camera is described, the form of the intrinsic and extrinsic camera parameters is showed. The relations between coordinates of a point in the world coordinate system with coordinates of this point on the image plane are given. The problem of finding extrinsic camera parameters from corresponded world and image points is posed. The error of found extrinsic parameters is estimated by simulation for several configurations of calibration objects consisting of a variable number of points on different distances from the camera and with different angles between them. On the basis of simulation results recommendations are given for the choice of the calibration object in order to improve the accuracy of the extrinsic camera parameters. The best results may be obtained with a configuration consisting of points that have a large angular distance between them and are posed on different distance from camera.

Keywords: camera calibration, extrinsic camera parameters, position and orientation estimation, calibration object

The short review of existing panoramic systems are given. The limits of such systems doesn't allow to form a panoramic video with dealing the parallax effect in real-time systems. The presented algorithm for forming a panoramic video is free of these disadvantages. An algorithm based on mathematical model of the round-view system that takes into account the location of cameras, their intrinsic parameters and model of the world. The algorithm includes following chain of coordinate transformation: from panoramic image coordinates to spherical coordinates and to Cartesian world coordinate system, from world coordinates to camera coordinate system and to input image coordinates. Geometric distortion of camera is dealt. Model of the world is represented as spherical and a hemispherical surfaces.

Keywords: panoramic image, round-view video system, camera model, camera distortion, parallax