Research And Application Of Binocular Ranging Algorithm Based On Visualization Technology

In recent years,with the continuous theoretical and practical breakthroughs in computer and image processing technology,computer vision technology has been widely used in many fields such as robot control,unmanned driving,non-contact measurement,and aerospace detection.Therefore,computer vision technology has become a domestic Research hotspots in foreign universities and research institutes.Binocular distance measurement algorithm is an important branch of computer vision field,which is mainly based on the principle of parallax and using triangulation to obtain the depth information of space objects.Binocular distance measurement algorithm is an important branch of computer vision field,which is mainly based on the principle of disparity and using triangulation to obtain the depth information of space objects.Because of its non-contact measurement and simple implementation,it can be used to solve the ranging problem in engineering.An analysis of the existing binocular ranging algorithms reveals the following three main problems:(1)The current binocular ranging algorithm is mainly used to measure the distance between a space point and the camera.In engineering,it is often necessary to measure the distance between any two points in space.(2)Image matching is the key to binocular ranging.The accuracy of disparity extraction determines the accuracy of ranging results,while the existing matching algorithms have poor accuracy and cannot meet the engineering ranging requirements.(3)In the fixed-focus binocular ranging algorithm,the measurement range is generally 0-10 m.When performing long-distance ranging,the ranging error is large,and the original binocular ranging algorithm is no longer applicable.This article focuses on the above three issues.The binocular ranging algorithm has been researched and improved.The main contents are as follows:(1)The binocular ranging algorithm is mainly composed of camera calibration,image correction,image matching,and triangulation algorithms.In order to measure the distance between any two points in space,this paper improves the original binocular ranging algorithm.In this paper,the three-dimensional coordinates of the space point relative to the left camera are obtained through mathematical derivation,so that the distance between the space points can be obtained using the distance formula.(2)This paper improves the existing template matching algorithm and proposes a differentiated image matching algorithm.The differential algorithm improves the contribution of low gray values to the image,reduces the impact of high gray values on the image,and makes the details of the image more abundant.This algorithm improves the correlation coefficient of image matching and obtains more accurate disparity value.(3)When performing long-distance ranging,the camera resolution is limited,and the disparity value of the spatial point in the left and right images is often less than one pixel unit,so the original binocular ranging algorithm cannot extract the disparity value.In this paper,the super-resolution reconstruction algorithm is used to improve it.By increasing the resolution of the image,accurate disparity values are obtained to achieve long-distance ranging.(4)This article builds a software and hardware platform,and uses a binocular algorithm based on super-resolution to conduct actual ranging experiments.By measuring objects in different scenes and different distances,it is found that this algorithm greatly improves the accuracy of long-distance ranging.