The Moving Target Tracking Technology Research Based On Binocular Vision

Tracking of moving objects has always been an important research direction of computer vision.After years of development,remarkable results have been achieved.However,there are still some problems in the current situation,such as the effects of tracking are not ideal when background disturbances,target deformations,target occlusion,and light changes.Taking into account the binocular vision can measure the three-dimensional information of the target,and it has strong anti-environmental interference ability.This paper studies the moving target tracking technology based on binocular vision,mainly including binocular camera calibration,binocular stereo matching,moving target detection and There are four parts of the tracking of sports goals.The specific research content is as follows:First of all,this paper studies the calibration method of binocular camera: explains the calibration principle of the camera,establishes the nonlinear model of the camera,and analyzes the conversion relationship of the four reference coordinate systems.The calibrating method was classified,and the flexible,fast and precise Zhang calibration method was selected for calibration.Through experiments,the camera's internal and external parameters were solved,and the stereo images were corrected for the left and right images.Then,the binocular stereo matching algorithm is studied: the binocular matching constraints are listed,and the stereo matching method is classified according to the selection of matching primitives and the difference of matching strategies.Three kinds of algorithms with low false matching rate are selected to study,and three matching algorithms are compared through standard test cases and real scenes.Experiments show that SGM semi-global stereo matching algorithm has good matching effect and high precision.This method is used to complete the stereo matching link and generate a disparity map,which lays a good foundation for the next target detection.Then,the moving target detection algorithm is studied.The traditional target detection algorithm is enumerated.According to the shortcomings of the traditional method in the misunderstanding of the complex environment,a moving target detection algorithm based on binocular vision is adopted,and the implementation flow of the algorithm is illustrated..Aiming at the similar gray levels,illumination changes and shadow problems in real-life scenes,the traditional method and the algorithm are compared and analyzed experimentally.The experimental results show that the proposed algorithm can overcome complex environmental disturbances and detect moving targets accurately.Finally,the moving target tracking algorithm is studied: the target tracking algorithm is classified,and a particle filter method with high applicability is selected for research.The particle filter algorithm principle and implementation process are analyzed.In order to solve the problems of background interference,single feature and low particle efficiency in traditional particle filter methods,particle filter target tracking algorithm based on binocular vision and multi-feature fusion is proposed.The improved algorithm solves the problem that the tracking effect is not ideal when the target occlusion,rapid movement and multi-target intersection are present.Through comparison experiments with traditional algorithms,it is verified that the improved algorithm can track the target accurately in real time.