Research On The Calibration And Positioning Method Of Vision System Based On Medical Collaborative Robot

With the development of robotics,the use of machine vision technology to detect and identify work targets and determine the location of targets can improve the intelligence of production operations.In the medical industry,medical instrumentation(medical collaborative robots)are also often used to work on the human body,such as treating stones,MRI,and analyzing blood cells.Combining medical collaborative robots with machine vision to make medical devices more intelligent has become a problem worth studying.In this paper,the calibration and localization method of recognition target is studied in two states,static and motion,using 2D code image as recognition means in the context of medical collaborative robots assisting doctors in surgical operations in medical devices.Firstly,this paper determines the internal and external parameters of the camera by using the Zhang Zhengyou calibration method under the premise of analyzing the principle of camera calibration.For the target object 2D code,character encoding is used to encode the 2D code image,and the pre-processing of the 2D code image is completed by using image binarization,image denoising,edge detection and morphological processing on the acquired 2D code image to obtain a target image with more complete information.Next,the static 2D codes are calibrated and localized.For the problem of low accuracy of ORB algorithm for image feature matching,this paper proposes a method to improve ORB algorithm by adding 2-nearest neighbor algorithm,two-way matching cross-filtering algorithm and RANSAC iteration,and the improved ORB algorithm can reach more than 90% accuracy for image feature matching.For the localization of static targets,the coarse localization of 2D codes is achieved by the estimated singleresponse matrix.Then,based on this,the fine localization of the target is completed by static experiments.Again,the dynamic 2D codes are calibrated and localized.The improved ORB algorithm first extracts the feature point pairs of the target 2D code,and for the problem of single color model when the Cam Shift algorithm tracks and calibrates the dynamic target,this paper adaptively fuses the H-component color histogram and the gradient direction histogram,and proposes an improved Cam Shift calibration method based on QR code position detection to realize the real-time tracking and positioning of the dynamic 2D code.The two Cam Shift algorithms before and after optimization are compared and experimentally verified.The improved Cam Shift algorithm has greatly improved the real-time,efficiency and accuracy for dynamic target calibration and localization.Finally,the medical collaborative robot vision platform is designed by MATLAB robotics toolbox,and the simulation experiments are designed in Simulink simulation software based on position vision servo control for static and dynamic target calibration and localization.Then,the target calibration and localization system of machine vision was designed and developed by Open CV computer vision library and VS 2017 software development environment,through which the static and dynamic targets of2 D code images were found accurately,and the applicability and effectiveness of the calibration and localization method of vision system proposed in this paper were verified.