Study On High-precision Calibration Of Digital Close-Range Stereo Photogrammetry System

Reconstructing the three-dimensional model of the world need a great amount of spacial data of objects, where fast and largely accessing high-precision point data is the basis of three-dimensional reconstruction. Compared with many other ways, using low-cast and efficient close-range photogrammetry system, which has gotten attention, to get points’informations of objects is good. However, lack of precision in results restricted it. The factors affecting the accuracy of results include the accuracy of control points, the applicability of the model and the correctness of the calibration of lens distortion etc.This paper studies the method of camera calibration based on high-precision three-dimensional calibration field. Some CCD cameras belonging to vehicle system and an ordinary camera used as collection devices has been used to shot to get close-range images, and camera calibration methods are studied and results of calculation are analysed. The main contents include that:1) camera calibration model existing before based on the calibration field have been studied and some suggestions could improve the accuracy have been proposed, and the integrated calibration model about intrinsic parameters and elements of exterior orientation of camera and the distortion model are built, and a method which is to iterate the two step which are solving calibration parameters and solving distortion parameters for improving the accuracy is proposed.2) According to the calibration model and distortion model, some formula for solving image points coordinate and objects’ coordinate based on intrinsic parameters and elements of exterior orientation of camera are deduced.3)The methods are implement by C++, and experiments is used to test and verify the method of calibration’s and solving both objects’ coordinate and image points’coordinate. And then the error ellipses of object points’ coordinate solved by images are drawn.The main achievement of this paper is that the integrated calibration model about intrinsic parameters and elements of exterior orientation of camera are built and a solution method associated with the model is implemented, solution accuracy about image points’ coordinate and object points’ coordinate is advanced obviously. The innovation in this paper includes that:1) the intrinsic parameters and elements of exterior orientation of camera are brought in one matrix, so that a similar rotation matrix implied parameters could be structured.2) The iterative process on similar rotation matrix and distortion parameters advances the accuracy of solution of parameters, image points’coordinate and object points’coordinate.Experiment results show that the method proposed is stable, and the root-mean-square error of resolution of images shot by CCD cameras are within1.3‰, and by ordinary camera is2.6‰. From this, it can be seen that the accuracy is better than the improved Weng method and DLT and non-linear iterative calibration three-stage algorithm, and not only CCD cameras belonging to vehicle system but also ordinary cameras can get a high-precision measurement of distance between two object points.