Research Of Camera Calibration And Stereo Matching In Binocular Vision

Computer vision is an important branch of computer science and artificial intelligence.From the1980s,Computer vision develops rapidly for the born of Marrr vision, Camera self-calibration and3D reconstruction for Nearly Three Decades. Binocular vision as an important part of computer vision, has become a front research projects in computer vision domain, whick is widely used in military and civilian field.Binocular vision system is a multi-disciplinary comprehensive technology, involving optical,image processing,computer sensor,artificial intelligence and electronic technology. The key and difficult technology in binocular vision is camera calibration and stereo match. Aim at the two research focus, the main research contents are as follows in this paper:(1) Camera calibration method based on planeCamera calibrate is a significant technique in computer vision,which aim at figuring out the relationship of the pixel poison and the3D position,finally get the intrinsic parameters(the internal camera geometric and optical characteristics) and extrinsic parameters (the3D position and orientation of the camera system relative to a certain world coordinate system). The paper gives a kind of improved plane calibration algorithm on the basis of analysising the traditional camera calibration. The algorithm has improved the precision of the calibration by using a graphic board as calibration references, in which the corner board accurate to subpixel precision position.(2) Improved stereo matching algorithmsStereo matching is the most difficult thing in binocular vision system, whose purpose is to get the3D information of objects through the2D information in image. This article proposes a new adaptive window stereo matching algorithms in view of the difficuty to choose support window in the traditional regional matching algorithm. The algorithm calculate the pixel gradient by using Sobel gradient operator, and dynamicly get the size of adaptive support windows according to the gradient value.At the end of this dissertation, the main research is summarized. It makes out the main research results,also give the shortage in the dissertation and the problem need to further research.