Research And Application Of Improved Binocular Stereo Vision In Ranging System

With the continuous development of industrial technology,the research of machine vision has made great progress.The research of binocular stereo vision has always been the focus of machine vision.Due to the continuous improvement of various algorithms in recent years,binocular stereo vision has greatly improved the matching effect,accuracy and speed.At present,binocular stereo vision has been widely used in UAV patrol inspection and other workTaking the perception of UAV to the surrounding environment as the starting point,this thesis deeply studies the stereo matching algorithm and ranging technology of binocular stereo vision,mainly including the following aspects: the basic calculation principle of binocular stereo vision,the calibration of camera parameters,image feature extraction and matching,distance measurement and so on.In order to obtain more accurate results in the process of camera calibration,Susan and Harris corner detection algorithms are studied in detail.After comparative analysis,Harris corner detection algorithm is finally selected.The matching corner information is obtained by Harris corner detection algorithm,and then the sub-pixel corner detection algorithm is used to obtain more accurate corner information.The sub-pixel corner detection algorithm uses the gray gradient method to extract corners and optimize them,and uses the least square method to calculate new corners.The experimental results show that this method improves the accuracy of camera calibration.In order to improve the matching efficiency in the stereo matching process and enhance the anti-interference ability to the environment,an improved ad census transform algorithm is proposed in this thesis.Firstly,in the process of matching cost calculation,the fixed pixels in the window are selected for calculation;In the process of cost aggregation,the dynamic programming of multiple paths is selected;In the process of disparity map optimization,multiple detection methods are used to improve the quality of disparity map.Finally,through the experimental part,it is concluded that the improved algorithm proposed in this thesis has great advantages in stereo matching efficiency.Finally,the software and hardware platform of the ranging system is built to measure the distance of the same object in normal light,strong light and weak light.It can be seen from the experimental data that the algorithm used in this thesis can improve the ranging accuracy and enhance the anti-interference ability when the environment changes.It has a wide application prospect in practical engineering projects.