Study On Recognition And Location Of Workpiece And Grabbing Based On Robotic Active Vision

In order to adapt the application of machine vision in industry,this paper studies workpiece recognition and location based on machine vision,and workpiece grabbing.The research is carried out by three aspects as follows:firstly,calibration of robotic vision system,including calibration of camera and hand-eye relations;secondly,image processing of workpiece recognition and location;thirdly,implementation of robotic vision system.The research work in detail is as follows:Firstly,calibration of robotic vision system:1.Camera calibration:According to the imaging principle of camera,we obtain a number of images of calibration target from different direction angles,then complete the calibration of camera internal parameters by MATLAB calibration toolbox.2.Hand-eye calibration of robotic active vision:In order to adapt the Googol GRB-400 robot used in this paper,firstly,through accurate controlling the manipulator to do two specific translation motions,the hand-eye relation rotation matrix can be calibrated with the points in scene;secondly,obtain the world coordinates of feature points on the workpiece platform by laser pointer,which is mounted on the mechanical arm,and the translation vector of hand-eye relations can be solved by the coordinate transformation relationship of robotic vision system;finally,the experimental result of hand-eye calibration is analyzed,which shows that the accuracy of the calibration method presented in this paper can meet the requirement of robotic workpiece location and automatic grabbing.Secondly,workpiece recognition and location:Image processing including image preprocessing.contour extraction,template matching based on Hu moment invariants,and obtainment of workpiece position and direction,is studied in this paper.In the aspect of workpiece image preprocessing,algorithms of grey image transformation,image filtering and threshold segmentation,and corresponding processing results are analyzed.Method of workpiece external contour extraction and extracting procedure are put forward,at the same time,the extracting results are analyzed.In regard to recognition and location of workpiece image,it is implemented by the method of template matching,through calculating the Hu moment invariants of workpiece contour,at the same time,obtaining method of workpiece position and direction is presented.At last,the result analysis of experiment shows that the presented methods in this paper are efficient and feasible.Thirdly,implementation of robotic vision system:Use Visual C++2010 as the system development environment.C++ programming is completed with the algorithms library of image processing provided by OpenCV.The function modules of system are designed as various classes for programming.According to system functions,including system settings,information display of system status and image,automatic recognition and location of workpiece and grabbing operation,other auxiliary functions and so on,system interface and system functions are implemented by programming.Finally,the experimental verification is carried out.