Research Of Camera Calibration And Feature Point Matching In Binocular Stereo Vision

Computer vision as an important branch of computer science and artificial intelligence has rapidly developed since the nearly30years. And binocular stereo vision is an extremely important part of computer vision. Binocular stereo vision can simulate human eyes for three-dimensional perception of three-dimensional objects, which is widely used in various fields of industrial automation and real life.Binocular stereo vision technology is a technology that comprehensive is strong. The knowledge is involved in many fields, including physics, optics, image processing, computer, artificial intelligence and electronic technology content. Camera calibration and feature point matching is the key and difficulty point in binocular stereo vision. In this article, aiming at these two aspects, the main research includes the following contents:(1) Camera calibrationCamera calibration is the foundation of the binocular stereo vision technology which has been applied, and also is one of the key technologies in the field of computer vision. The purpose of calibration is to establish the relationship between the position of the pixel of the camera image and the position of the three-dimensional target object, and ultimately determine the camera intrinsic parameters and extrinsic parameters. The methods used in this article:combining with OpenCV programming online calibration of the internal parameters of the camera and using of essential matrix inverse solution to obtain the camera external parameters of binocular stereo vision system.(2) Feature point matchingFeature point matching is the basis of subsequent3D reconstruction and other technologies that can be come true. And it is also the most difficult part of the binocular stereo vision system technology. The purpose of the matching is to find in the image on the left and right image points of the same name, and the parallax calculated by some means. In this article, we use the improvement of the existing Harris feature point detection algorithm, which achieve the corner detection and enhance the ability of dealing with the image feature points, then combine with program to get the sub-pixel coordinates of corner, use SIFT feature descriptor to describe the feature points, finally put forward a matching algorithm which is based on the Harris feature points and the SIFT feature descriptor. And then we can obtain the disparity map after the matching.The article finally has carried on the summary to the full text, expounds the main research results, and puts forward some algorithm of this article and further improvements.