Stereo Vision System And Its Application In Measurement Of Static Size And Dynamic Trajectory

With the rapid development of modern manufacturing,the demand for automation and:intelligence becomes higher and higher.Machine vision with its non-contact,high accuracy,continuity,high speed,high flexibility,abundant information and other advantages,has great theoretical significance and practical value in the field of quality inspection,mechanical processing,target tracking.Based on the principle of stereo vision,the camera calibration,image acquisition,preprocessing,feature extraction and matching,measurement and other key technical problems were studied.The hardware and software platform for vision measurement were bulit for static dimensional measurement and dynamic trace detection.The main research includes:Introduce the background and significance and development status and trend of vision measurement technology.The hardware performance indicators and selection basis in vision system was analysed.The whole hardware platform was bulid with Basler industrial cameras and Kowa lens and STM32 and other vision hardware.The STM32 development board was used to trigger cameras for synchronous acquisition.The camera model was analysed.The ideal vision model and actual distortion model were established.The calibration of Zhang was studied.Single canera calibraition,binocular calibration,distortion correction were studied.The theoretical basis and related calculation method of the conversion from image coordinate to real size coordinate was taken by principle of 2D and 3D vision measurement.Vision measurement involves the analysis of image filtering,binarization,geometric characters detection.Contrasting Hough algorithm and least square algorithm,roundness error and least square method were combined to detect the circular feature of workpiece.Smallest external rectangle and harris algorithm were combined to get the sub-pixel coordinates of vertex points of rectangular feature for 2D vision measurement.Based on the harris-ncc algorithm,corner feature was detected and matched.The match was improved by further optimization.The developed stereo vision system was applied to the machine vision experiments of dimensional measurement about workpiece and three-dimentsional trace detection of badminton.The error was analysed.The performance parameters was cleared.The proposed algorithm was proved to be useful.At the same time,it also provided a solution for some advanced stereo vision algorithms in the practical field of industrial static measurement and dynamic target detection.