Study On Recognition And Localization Of Stacked Parts Based On Binocular Vision

With the development of signal processing technology and computer technology,computer vision has been applied in more and more occasions to accomplish tasks that human eyes cannot do.In the field of industrial production,industrial manipulators need stereovision to play a guiding role in the process of grabbing workpieces and automatic sorting.At this time,due to the disordered orientation of workpieces on the worktable or on the conveyor belt plane,and the situation that two or more workpieces cross and are stacked together,the workpieces in this case are different from a single workpiece lying on the worktable or on the conveyor belt plane.In this case,the difficulty of grabbing by industrial manipulators will be increased.Solving the problem of identification and positioning of stacked parts is of great practical significance for accurately grabbing and sorting workpieces in production.In this paper,the recognition and positioning of stacked parts based on binocular stereo vision system are deeply studied.The specific research contents are as follows:Firstly,the research background and significance of the recognition and positioning of stacked parts based on binocular vision are described,the current research status of binocular vision technology at home and abroad is analyzed,the shortcomings of binocular vision in the recognition and positioning of stacked parts are summarized,and the "HALCON" software applied in the implementation algorithm is briefly introduced.Secondly,the overall scheme of binocular vision recognition system for stacked parts is designed,including hardware module and software module design.Debugging experiments are carried out on the software design scheme,and relevant operators are edited in the HDevelop program editor,thus realizing the connection process of software and hardware.Thirdly,the binocular vision calibration algorithm applied in the position and orientation recognition and positioning of stacked parts is studied.Specifically,a camera calibration method based on HALCON binocular vision is proposed,the calibration principle and calibration process are verified,and the calibration results of internal and external parameters of the camera are obtained.Fourthly,the contour feature extraction method of image segmentation stacked parts based on Blob analysis and the template matching method based on contour feature are applied to complete the identification and positioning process of parts.taking stacked small threaded rod parts as an example,the experimental verification shows that the method has better rotation invariance and scale invariance.Finally,in the stereo matching algorithm,the NCC stereo matching algorithm based on optimization is proposed,and the stereo matching of stacked parts is realized on HALCON software,and the matching effect and result are given.At the same time,the recognition and positioning system of stacked parts is realized by using C# language.