Research On Target Localization Algorithm Of Binocular Fisheye Vision System Under Moving Base

Computer vision technology originated in the 20 th century and developed rapidly in the 21 st century.Based on computer vision technology,target recognition technology,target positioning technology,target tracking technology,etc.have also continuously achieved new breakthroughs and applications.Among them,the binocular stereo vision positioning technology system is widely used in many fields such as life service,industrial knowing,national defense construction,etc.,with the advantages of simple structure,no contact work,low cost,good real-time performance and high precision.Taking the ocean monitoring in national defense construction as an example,the binocular stereo vision positioning system installed on the monitoring ship should have the ability to accurately locate the air target.However,the current general binocular positioning system uses a common camera,and its field of view angle is small,which cannot meet the requirement of positioning under a large field of view for sea-to-air detection.Moreover,the situation at sea is complicated.Due to the presence of the waves,the monitoring ship will oscillate from time to time,and the binocular stereo vision positioning system fixed on the hull will also follow the ship's sway.Usually,the reference coordinate system we establish is based on the camera coordinate system of the left camera.This kind of sway will cause the relative positional relationship between the camera coordinate system and the sea level coordinate system to change.Eventually,the target motion trajectory we draw will no longer be the true motion trajectory of the target observed when people are still.Based on the above description.Aiming at the problems of small field of view and swaying of the common binocular vision positioning system,this paper studies the stereo vision positioning algorithm of binocular fisheye under the moving base.Firstly,this paper introduces the principle of binocular stereo vision positioning and determines the parameters of binocular positioning system.In the first part,the paper introduces the conversion relationship between coordinate systems and coordinate systems commonly used in vision systems.The understanding of the coordinate system is the premise of studying the binocular positioning model and deducing the ideal positioning formula.On this basis,the accuracy of the target positioning is further discussed according to the ideal positioning formula.The positioning accuracy is an important condition for determining the parameters of the binocular stereo vision positioning system.At the same time,the first part also discusses the visual blind zone of the binocular system,and the comprehensive positioning accuracy and visual blind zone determine the equipment parameters of the fisheye binocular stereo vision positioning system.Secondly,this paper studies the image distortion correction algorithm under large field of view.In this section,this article introduces camera calibration techniques,including Zhang Zhengyou calibration and five-point calibration.Camera calibration technology is the premise to determine the relative positional relationship between binoculars,image dedistortion and stereo matching.In this paper,Zhang Zhengyou's calibration method is used for single target determination,and the five-point calibration method is used for double target determination.At the same time,the five-point calibration method is improved to make it possible to determine a unique solution when there are multiple basic matrices.Further,this paper introduces the distortion model of ordinary cameras and fisheye cameras,and related stereo correction algorithms.On this basis,this paper corrects the image pair and gives a correction map.Thirdly,this paper studies the large-field stereo vision positioning technology and tilt compensation.This is the last part of the theoretical study of this paper.In this part,three typical stereo matching algorithms are introduced,and the SGBM algorithm is used for stereo matching,and then the disparity map is given.At the same time,this paper proposes a new tilt angle compensation algorithm.When the base of the binocular fisheye positioning device oscillates,the tilt angle compensation algorithm is used to compensate the swaying binocular fisheye positioning system by the output angle of the tilt sensor.The deviation angle of the swing is compensated into the system,and the target coordinates in the current state are restored to the initial coordinate system,that is,the coordinate system of the three-dimensional coordinates of the target obtained finally is stationary with respect to the geodetic coordinate system or sea level,so as to conform to people's observation habits.Finally,this paper builds a fisheye binocular stereo vision positioning system on the swinging platform,and introduces the selection of relevant units in the equipment and the design of the computer control interface.In this paper,the positioning accuracy of the binocular fisheye positioning system is verified by the target positioning experiment under the static base.The compensation accuracy of the binocular fisheye positioning system is verified by the tilt angle compensation experiment.