Research On The Motion Analysis System For Magnetorheological Robot Based On Machine Vision

The magnetorheological robot motion analysis system based on the machine vision is a system for measuring the position and speed of a magnetic robot,based on computer vision.The use of non-contact optical measurement methods greatly simplifies it.The development of registration equipment,on the one hand,and the methods of object tracking(computer vision section),on the other hand,open the possibility of widespread use of optical measurement techniques in experimental studies of systems with sophisticated computer vision.Object tracking is an active direction of computer vision.There are many different methods for tracking objects in a series of images,but most of them are unsuitable for an accurate natural experiment.In primary applications,computer vision,such as security video surveillance,traffic flow analysis,does not have to be very accurate in determining coordinates.The methods are mainly aimed at solving several related problems.Highlighting an object from a dynamic background in variable lighting conditions Improving image quality,processing data in real-time.The natural physical experiment,on the one hand,imposes increased requirements for the precise definition of the coordinates of objects;on the other hand,in the experiment,it is possible to provide better images and consistency of lighting.This avoids additional image processing and focuses on the main task-the exact definition of coordinates.In this regard,studies have begun to adapt existing methods of computer vision to natural mechanical experiments,as well as the development of new methods.This research focuses on the study of computer vision techniques to measure the position and speed of a magnetic robot and includes the following aspects:(1)The basic theory of optics,types of cameras and lenses are the main factors influencing the caste to images such as lens distortion,and the vignette effect.The camera interfaces were also reviewed,and their bandwidth and other technical limitations were compared.On the basis of the theory of photogrammetry,the camera calibration algorithms were considered.The internal and external characteristics of the camera,the characteristics of the lighting,and the methods of calibration of the camera to measure the speed have been defined.The design of the magnetorheological robot motion analysis system has been chosen.(2)The basic theory of pre-processing of images methods of improving images has been introduced,the main factors influencing the quality of the image have been identified.The main types of noises and ways to suppress them have been identified.The use of basic imaging techniques such as histogram image processing,Fourier conversion,and the effect of frequency filtering on data extraction from the image was examined.The basic methods of segmentation of images for the identification of objects in the image according to histogram analysis were considered.The methods of the nearest neighbors and the Hough transformation to determine geometric primitives are considered.(3).The methods of tracking objects on the sequence of images were considered,the main shortcomings of tracking methods were identified,and the accuracy of the position and speed of operation were compared.Tracking linear dynamical motion was also studied to track the movement of the magnetorheological robot at a constant speed.As a result of the research,it was determined that the most accurate and fastest method of tracking the object is a contour method based on the Hough transformation,the accuracy of which was 0.01 pix.(4)A test stand of the magnetic robot's motion analysis system has been installed.The camera was calibrated to determine a static error.The software part used the opensource program is Lab View environment using NI Vision Builder for Automated Inspection and Lab VIEW Vision Development Module,Vision acquisition software that includes the IMAQ drivers to work with video cameras in real-time.The functionality of the software allows track elements using all methods of computer vision.The experiment used the Hough transformation method for the segmentation of images and further tracking of the object.There are 52 figures,14 tables,and 90 references in this paper.