The 3-D Reconstriction And Defect Detection Of Module Based On Digital-image

Three-dimensional reconstruction technology is an important subject areas in machine vision, contributing to the achievement of non-contact measurement and three-dimensional reconstruction rapid modeling. The technical is high efficiency, is favourable to improve the automation of the production,is an important direction of development 3D modeling.This paper detailedly research camera calibration, feature extraction three-dimensional matching in binocular stereo vision reconstruction,and so on.Zhang's calibration methods is used in camera calibration.The methods used grid template as the camera calibration plate, calibrate the cameras in the two locations and access to the camera's internal and external parameters.After shooted the same workpiece (cube workpiece) by the camera in two directions and processed,we respectively extract feature points in two images.secondly we find matching feature pointsin two images based on epipolar constraint, and other methods.Finally we access the corresponding space coordinates of match point by calculating. Through the analysis and comparison,it is selected 2D median filtering, Canny edge detection methods and extraction methods corner. In the analysis of various image preprocessing algorithms,we achieved these algorithms On the bace of VC + + and Matlab mixed programming method.We completed the entire process of reconstruction Based on the stereo matching of feature points.On the basis of two cameras inside and outside the parameters and matching feature points,we calculat the three-dimensional spatial coordinates of points.finally it is Reproduced in the OpenGL 3D to objects feature points form.