Design And Research Of Spherical Valve Body Recognition And Positioning System Based On Stereo Vision

With the rapid development of society,human demand for materials is also increasing rapidly,resulting in new challenges and new forms of intelligent manufacturing.In order to meet the needs of human life and adapt to the new market situation,enterprises improve work efficiency and reduce costs by using robots instead of manpower on the basis of considering cost,safety and efficiency.The traditional industrial robot can only work in the teaching or off-line programming mode.In this mode,the robot trajectory is fixed and does not realize the automatic operation of the robot.At present,the combination of vision and industrial robot is of great significance to improve the working efficiency of robot and realize automatic production.In this paper,binocular vision is applied to industrial robot to make the robot have the ability to perceive the changes of surrounding environment.The research work is carried out on the recognition,positioning and grasping of spherical valve body.Taking the built robot as the experimental platform,aiming at the inconvenient traditional teaching and off-line programming of the current industrial robot,as well as the time-consuming and labor-consuming problem of the spherical valve body in the handling process,a set of spherical valve body recognition and positioning system based on stereo vision is designed by using MFC,C++ and Open CV.The specific research contents can be summarized as follows:(1)Design the overall scheme of the system according to the requirements of the subject task.Calibrate the selected DFK33G274 industrial camera,obtain the internal and external parameters of the camera,and study the binocular vision positioning problem.After comparing and analyzing the results of two camera calibration methods of Matlab and Open CV,Matlab toolbox is selected to realize camera calibration.(2)Identify and position the spherical valve body.Firstly,histogram and bilateral filtering algorithm are selected to enhance the image.In the recognition stage,this paper proposes a SIFT image matching method based on quadtree.The feasibility of the algorithm is verified by self built data set and standard data set.In the stage of valve body positioning,combining the matching algorithm proposed in this paper with the traditional Grab Cut algorithm will effectively solve the problem that Grab Cut needs to label the foreground manually.In the stereo matching stage,the template matching method under the constraint of stereo correction is adopted to search the points matching the center point of the left image in the right image to realize the three-dimensional reconstruction of the center point.(3)Build a spherical valve body recognition and positioning system based on stereo vision.Firstly,build the hardware platform,and then take Visual Studio 2015 as the development platform to realize the development of client application by using MFC,Open CV and C++.Finally,on the system,the identification,positioning and grasping experiments of the spherical valve body are carried out,and the accuracy error is experimentally analyzed.The experimental results show that the hardware platform built in this paper can communicate well with the programmed client application,and can meet the needs of spherical valve body recognition,positioning and grasping.