The Research Of Calibration Algorithm For Camera Internal Parameters Of 3D Coordinates Measurement In Large Space

With the development of signal processing theory, electronic technology and computer technology, the vision coordinate measurement technology has already been developed greatly in the last few years. Because of its advantages of non-contact, real-time, automatization and so on, the vision coordinate measurement technology has been widely used in national economy, science research and industry and is highly emphasized by many fields all over the world. Large-scale computer vision coordinate measurement which has been developed in recent years is one kind of vision coordinate measurement. Industrial applications of the large-scale computer vision coordinate measurement system cover a wide range from dimensional verification and checking to alignment, 3D documentation, displacement monitoring and deformation measurement, as well as numerous other 3D measurement tasks. In the process of measurement, it concerns several steps such as camera parameters calibration, feature points recognition, feature points matching, 3D reconstruction and so on. And internal parameter calibration is the first and the most important thing we need to do, because its accuracy determines the precision of the whole measurement system directly.There have been many camera internal parameters calibration methods presently, including independent calibration method, plane calibration method, two-plane calibration method, stereo calibration method, vanishing points calibration method, self calibration method and so on. But these methods do not consider the information on the depth direction because of the limit of target size, so the internal parameters can not be calibrated fully and the calibrating precision is low. Therefore, these methods can only be used in small-scale calibration occasion. For large space 3D coordinate measurement system, the foreign companies such as Metronor, Leica and Krypton adopt special calibration methods, but they are all kept secret.For the requirement for calibrating precision of large space 3D coordinate measurement system, this paper adopts virtual stereo calibration method based on the analysis of all kinds of calibration methods. For virtual stereo calibration method, the virtual stereo target in the 3D space is formed by a feature point which is taken by the 3D coordinate measurement machine, then the target is imaged in several stations by the calibrated camera, and finally the accurate camera internal parameters can be got by using special optimizing solving algorithm. Because the space information is considered at all directions, this method can get high calibrating precision in large space and can be used in large space 3D coordinate measurement system and realize single-camera, double-camera and multiple-camera measurement.