The Realization Of Key Technology In Stereo Image Three-dimensional Reconstruction Based On OpenCV

Our world is three-dimensional, but camera image is two-dimensional, in order to obtainthe three-dimensional information of objective object’s by the two-dimensional images, weneed to rebuild the three-dimensional model of the target so that we can target the quantitativeanalysis. The same subject in computer vision and photogrammetry community,the stereovision-based three-dimensional reconstruction techniques have been widely used in medicalimaging, robot navigation, virtual reality, terrain exploration and other fields.The3D reconstruction’s key technologies based on stereo vision include cameracalibration, feature extraction and stereo matching, determine depth information and the3Dcoordinates calculation. Against the problem of camera calibration in stereo vision, on thebasis of analysis the camera calibration model based on OpenCV and the existing cameracalibration method, This article give a method which consider the camera lens distortion,based on the OpenCV stereo camera calibration. By using this article’s method and ZhangZhengyou Mat Lab Calibration Toolbox calibration comparison experiments on real data, theexperiments show that the calibration parameter’s results stability, the accuracy can meet therequirements of stereo vision3D reconstruction and other applications.Direct the problem of feature extraction and matching in three-dimensionalreconstruction, this paper expounds the epipolar constraint of matching criteria, put matchingconstraints in a pole line, effectively removing the error matching and greatly improve theefficiency of matching. This paper use the latest research achievements in image processing—SIFT feature extraction matching algorithm,then designed two experiments to test theperformance of the SIFT algorithm, experiments show that the algorithm can solve theproblem in stereo matching such as image rotation, zooming, panning, etc. It has a highmatching accuracy,it is a very good algorithm. Determine the depth information and the three-dimensional coordinate computing is akey step in the three-dimensional reconstruction,Through the analysis of three-dimensional camera model,this paper give a method tocalculate the three-dimensional coordinates of space point based on the basic principles ofstereo vision,and then design experiments to generate the disparity map and calculate thethree-dimensional coordinates of spatial points, Validation the calculation results based on theknown information and analysis the error, the experiment proved that this method is effective.OpenCV is a open source library for computer vision technology, which contains a seriesof key technologies involved in the three-dimensional reconstruction module, researchers canquickly build their own platform in OpenCV to work. In this paper the experimentalalgorithm is programmed in VC++and OpenCV environment.