A Research On The Key Technology Of A Photogrammetry And 3D Reconstruction System

In recent years, more concern has been given to 3D reconstructing technology. This thesis introduces the essential principle of digital photogrammetry and 3D reconstruction briefly, then a method based on relative movement for camera calibration is brought up, and it is compared with other existing algorithms. After that a method based on control-point restriction for feature matching is given. Finally an integrated system of 3D reconstruction designed and implemented by us is introduced, and experiment results are given.Conventional methods for camera calibration usually need special calibrating-objects. While self-calibration methods do not, they extract camera parameters from images directly. Each kind of method has its strongpoint and limitation. For our project, we have come up with a method for camera calibration based on relative movement. It divides camera parameters into several groups and calculates them separately. Experiment results show that it makes answers nicety and rapidly, and its requirement for instruments is low.Existing algorithms for feature matching of image are mainly based on area, feature or hierarchy. But there is no algorithm can do it fully automatically. In this thesis a new method for feature matching is given. It matches several control points, and then matches other points based on them. Through experience we found it effective.Our 3D reconstructing system includes several clients and a server. The clients collect, arrange and send back information. The server manages the data and implements the work of 3D reconstruction. My work is mainly about the software in the server, including management of original database, camera calibration, image correction (partly), 3D reconstruction, model adjustment, and so on. Necessary interface between human and computer is included of course. The most important parts are camera calibration, image correction and 3D reconstruction, and our methods have been applied in them. Experiment results show the system can achieve the task of 3D reconstruction properly.