Binocular stereo vision system is a branch of computer vision, is the process of getting thepoint3D information through shooting the same scene by the different positions of the cameramoving or rotating and computing the point disparity in the left and right image. The mainprocesses are image acquisition, camera calibration, image matching and3D reconstruction.The paper designed the binocular stereo vision system based on the stereo vision of animals,using two cameras imitate the eyes of animal to get the image scene information. Using themethod based on calibration board calibrates the binocular stereo vision system, through opticalanalysis for binocular stereo vision, a method of direct image mosaic was proposed, that is themethod which is no need to look for the same character in the two adjacent images to make suretheir relative position through the relevant algorithm, stitch the image sheared from the initialimage to the large view image, realizing large view image mosaic. Through computing thecorrelation at areas of mosaic for the dislocation of the adjacent image made by systematic error,to get the best matching position of image mosaic edge and get the dislocation value to correctthe image mosaic., Using improved weighted method eliminate seam-line produced by differentlight intensity between the cameras.Finally, the paper uses HALCON software which has powerful image processing functionsto calibrate the camera system, and gets three-dimensional coordinate of the mark point in theimage, realizes three-dimensional digital measurement and uses Delaunay Triangulation methodto fit3D surface. Then using Visual Studio2010and AutoCAD secondary development toolkit ObjectARX2013develop and design software, and using object oriented technique realizes3Dreconstruction. |