Research On 3D Reconstruction Based On The Binocular Disparity

The binocular dispartity belief 3D reconstruction technology is a hot research topic of the computer visual. it has important application value in real life. It can not only applied to the industrial test, biomedicine, virtual reality technology, the key fields may still be applied in aerospace telemetry, reconnaissance, etc.This paper firstly introduces the computer visual and binocular stereo vision and the research content and development situation.then, The problem for 3d reconstruction of each step makes a detailed study, Mainly include: the camera calibration, image collection, feature extraction, solid matching and 3D reconstruction. For each step of various algorithm and method of research and analysis, mainly do the following:First studied camera model and fundamental matrix, fundamental matrix is an important research content in computer visual, it provides the mathematical model for the camera calibration, the research on fundamental matrix for the camera calibration provides theory basis in the next work.The camera calibration work, making the camera calibration template,discuss various methods of the camera calibration, finally in the matlab environment, use Camera Calibration Toolbox of matlab provides the matlab tool for the Camera Calibration experiment, obtain the internal and external parameters for the camera,it do a premise for the 3D reconstruction. according to the calibration data,we form the configuration of the parallel double camera system, It can imitate human visual on the video scene.Image feature extraction,the image feature extraction has a important meaning of the premise in the 3D reconstruction,this paper has researched many algorithm of the image feature extraction,it has cleared obstacles for the stereo matching work.Stereo matching, stereo matching is the key of 3D reconstruction and also the most difficult step, Now researchers have put forward a lot of stereo matching algorithms, This paper first studied various typical of stereo matching algorithm, then puts forward an improved algorithm of Segment based matching, from the experiments can be informed that this algorithm is effective.Finally, based on the stereo visual model reconstruction of the 3D scene depth graph, and to achieve the expected effect of the algorithm.