Design And Implementation Of Camera Ranging System Based On Binocular Vision

As an indispensable part of the research field of computer vision,binocular stereo ranging technology is composed of many disciplines.The key step of binocular ranging is to obtain the left and right images by two cameras with identical specifications,and to obtain the location information of the target points after a series of information processing of each set of two-dimensional image pairs,the technology has the characteristics of simple implementation and low cost which has been widely used in some non-contact measurement fields.This paper designs and implements a camera ranging system based on binocular vision,the main work includes the following aspects:Firstly,four important coordinate systems in the camera imaging model are determined,and the binocular ranging model is derived through the transformation relationship between the four coordinate systems;the different calibration methods are analyzed and studied,camera calibration experiments are carried out by using Zhang Zhengyou calibration mode and Matlab toolbox,the internal and external parameters of cameras are obtained;in view of the phenomenon that the images are inconsistent due to different angles of view of the left and right camera,Hough correction is used to make the left and right images reach the state of strict line alignment;in the link of feature extraction,the advantages and disadvantages of different extraction algorithms are analyzed and studied,and the problem of high dimension of feature points and low resolution of SIFT algorithm is optimized,the traditional 128 dimension is reduced to 24 dimensions to calculate,which greatly reduces the running time of the algorithm,a large number of similar pixel points are distinguished by reducing the number of comparison times between the minimum priority queue,thus improving the resolution.The experimental results show that the improved SIFT algorithm optimizes the image quality.Secondly,the experimental results of SGBM algorithm,BM algorithm and GC algorithm are analyzed and compared,finally the SGBM algorithm is considered from the time and matching accuracy and the matching experiment is carried out.Aiming at the problems of insufficient accuracy and long running time of single SGBM algorithm,the matching points set of Sobel operator in the step are screened and the mismatching points are eliminated by fusion RANSAC algorithm,so that the final matching efficiency is improved and the position region of the target object is located.The experimental results show that the improved SGBM algorithm is more accurate in feature point matching.Finally,a camera ranging system based on binocular vision is developed by combining Opencv visual library and Matlab calibration toolbox in Microsoft Visual Studio development environment.The experimental results show that the system basically realizes the main functions of ranging experiment,that is,camera calibration and image correction,image preprocessing and feature extraction,stereo matching and target location,ranging.