Research On Detection And Location Of The Golf Based On Binocular Vision

As the binocular stereo vision system develops,more and more intelligent service robots are equipped with "eyes".However,vision system of pick-up robot still has backward technology.In order to solve the problems of picking up robot vision system and popularize the application of stereo vision system in service-oriented robot.In this paper,a recognition and location system of Golf based on binocular vision is proposed.This paper studies the difficult problems around stereo camera calibration,image preprocessing,sphere image matching,location information extraction and so on.The research includes the following aspects.(1)Camera calibration and location of stereo vision system.Firstly,in order to analyze the mapping relationship between spatial points and pixels,the coordinates of visual systems are established,and the relationship of mutual transformation is analyzed.The camera model system is established by the camera linear model and the camera nonlinear model.Furthermore,two methods of camera calibration(Direct calibration method and Zhang calibration method)are discussed according to the camera linear model.The results of calibration experiment are good.Finally,the principle of stereo location is analyzed to pave the groundwork for the analysis of location accuracy.(2)Image preprocessing and Golf detection.In order to obtain good recognition and detection,the image preprocessing mechanism is established.Firstly,the principle of image grayscale,binarization,morphology and Canny edge detection are analyzed in detail.The experimental results of real-scene images are also given.Finally,the principle of image detection based on Hoff transform is also analyzed.(3)Matching and location of Golf images.Using theory and experiment to analyze whether template matching and feature matching are suitable for Golf matching.Secondly,by analyzing the principle of epipolar geometry and the characteristics of Golf image in detail.A theory of Golf matching algorithm based on limit constraint is proposed.Experiments are used to verify the effectiveness of the algorithm.Finally,the spatial information of Golf is obtained by section 4.5.Many of methods is verified,for example the camera calibration,image preprocessing,sphere detection and extraction,Golf matching algorithm and others.(4)Stereo vision location accuracy analysis.In order to improve the location accuracy of binocular stereo vision system,the model of vision system is established,and the impact of structural parameters(baseline distance,camera focus,angle of between optical axis and baseline)on location accuracy is analyzed,as well as the impact of calibration parameters(camera distortion,number of calibration images,number of targets,position of calibration board)on location accuracy.The trend of parameters and the optimum range are given by simulation and experiment.It provides an important basis for the binocular stereo vision system to further research.To sum up,the research results of this paper can provide technical support for the vision module of Golf picking robot.On the other hand,it also provides technical reserve for vision detection and location of all spherical objects in the future.