3d Reconstruction Of Digital Camera Binocular Stereo Vision

Binocular stereo vision realizes the 3D reconstruction of objects by simulating human vision mechanism to get the deep information of image, which is a significant research topic in machine vision. Nowadays, with the constantly surge of digital camera, it has become a burning problem for common users to capture 2D scenery by digital camera and realize the 3D reconstruction of the scenery. This thesis accomplished a binocular stereo vision based 3D reconstruction system with the foundation of binocular stereo vision theory using common digital camera and learned the camera calibration as well as the stereo matching in 3D reconstruction. The main work is shown as follows:1. Theories and algorithms of camera calibration were studied and planar checkerboard calibration template was design. X-corner detection algorithm in the calibration process was modified. Finally, digital camera was calibrated and internal parameters were obtained.2. Stereo matching algorithm was studied. Harris based and SIFT based feature detection algorithms were compared. Finally, fundamental matrixes were got according to matching result and external parameters were calculated using fundamental matrixes and internal parameters.3. 3D reconstruction system was built by current conditions. 3D coordinates of feature points were calculated with stereo vision principle using the internal and external parameters of the camera as well as the stereo matching results. Eventually, these coordinates were shown with OpenGL and the 3D reconstruction was realized.