A Study On Matching Of Objects For Binocular Stereo Vision

Binocular stereo vision is a vision system to simulate human visual sense using a binocular camera.It obtains the two background images of the measured target area from two different locations at the same time and achieves three-dimensional information recovering by target parallax relationship in two background images.Binocular stereo vision measurement system has many advantages,such as simple structure,large measuring range and high measuring precision.It is very suitable for the research of three-dimensional localization of targets in large field of view.For practical demand of localization of multiple random objects with large view field,long distance and same features,a three-dimensional coordinate measuring system based on binocular stereo vision principle is established.In order to achieve the high-precision positioning of the multiple objects with same features,it is necessary to match the multiple objects with same features correctly at first,and then use the calibration model to solve three-dimensional coordinates,and finally achieve the positioning of multiple objects with same features.As a result,the three-dimensional calibration solution model is mature,and its accuracy is high.Therefore,this paper mainly deals with the stereo matching problem of the multiple objects with same features in binocular stereo vision.The method can be applied to the automatic evaluation system of the artillery firing.To solve the key problem of stereo matching of multiple objects with same features in binocular stereo vision,the main work of this paper is as follows:(1)The principle of stereo matching of multiple objects with same features is analyzed.The principles and methods of ASIFT algorithm,RANSAC algorithm,epipolar constraint,affine constraint,and Delaunay triangulation are described respectively.(2)The method of matching the multiple objects with same features based on GPS time frame synchronization and epipolar constraint is improved.Combining epipolar constraint and affine constraints,the matching of the multiple objects with same features is achieved by searching for the area of the intersection of the epipolar constraint and the affine constraint.The experiments show that the correct rate of the matching of the multiple objects with same features is improved by 15% to 20%.(3)A method for stereo matching of multiple objects with same features based on Delaunay triangulation and affine constraint is proposed.The experimental results show that the proposed method achieves fast and high precision matching for the objects with same features.It's successful to solve the problem of matching the multiple objects with same features on the arc-shaped slope in a large three dimensional space.