Binocular Image Computer Measurement System

Computer stereo vision is a new and developing technology with wide application prospect in the field of modern industry and 3D measurement. This paper uses binocular stereo system simulating the eyes of people and research on the key technology of stereo vision: camera calibration,image processing,characteristics points obtaining and matching,3D reconstruction et al. Binocular stereo vision measurement is also the most prevalent realize method basing vision measurement The main work and achievements of this thesis are as follows:In the stage of camera calibration, we introduce the newest 1D calibrate theory which was presented by Zhang Zhengyou, and give some improvements. Realize camera calibration and rectify the lens distortion using 2D calibrating plate.In the stage of image pre-processing, give a detail introduction and compare of the fitting methods. Realize characteristics points obtaining using Harris arithmetic. At last ,use Opencv library to realize fast characteristics points obtaining.In the stage of stereo matching, using the matching method based the character points, we do a lot of research on character points matching method and compare the effect caused by different operators and windows. At last, we use stereo matching method based on minimization of energy function and do some improvements making use of inconsistence restriction with pyramid matching strategy in order to get dense disparity images.In the stage of 3D reconstruction, based on the results of the camera calibration and character points matching, we reconstruct the character points according to the mathematical model of the space points, and get the 3D coordinates of the character points. At last, we achieve the 3D curve model using fitting.This paper uses VC++ and Matlab mixed programming with opencv library to realize the key steps.