Research On Visual Positioning Technology Of NC Machining

With the rapid development of industrial camera and image technology,machine vision technology has been widely used in various fields of industrial production,and has produced huge economic benefits.CNC machine tool processing before the workpiece positioning,the traditional contact positioning method is low precision,troublesome operation,and visual measurement positioning technology as a non-contact measurement technology,high detection accuracy,high efficiency,but also can avoid subjective error caused by artificial.Therefore,combined with the practical application requirements of CNC machining,this paper studies the calibration method of vision system and designs and realizes the relevant algorithm of rectangular workpiece positioning and measurement.The main research contents are as follows:The pinhole imaging camera model and the traditional camera calibration principle taking Zhang Zhengyou calibration method as an example are studied,and the Zhang Zhengyou calibration method is implemented with Open CV library.Aiming at the characteristics of rough calibration wiring strip and unclear image of machine tools,an image sharpening algorithm based on Sobel operator was designed,and the coordinate value of corner points was successfully extracted by subpixel detection algorithm based on Harris corner detection algorithm.In this paper,a camera calibration method based on radial basis function neural network is proposed.The RBF neural network class of C++ is designed,and the initialization,training,calculation and output functions of the neural network are realized.The image distortion correction scheme is designed.The pixel equivalent calibration method was studied,and the transformation algorithm between image coordinate system and machine tool coordinate system was designed to realize the visual automatic positioning function of NC machining.A morphology-based contour extraction scheme was proposed.The OSTU method was used to perform binary segmentation of the image,and then the contour map of the single edge of the object was extracted by etching and other operations.The eight-neighborhood tracking method was used to track the contour,and Douglas-Peucker algorithm was used to fit the contour polygon.Visual detection of shape and size of rectangular workpiece is realized.The hardware platform of rectangular workpiece visual positioning measurement system is built,and the corresponding software is designed and developed.MFC,Open CV and other tools were used to develop the visual measurement and positioning software system,which realized the functions of image correction and rectangular workpiece detection,positioning and measurement.Experimental results show that the reprojection error of the proposed calibration algorithm is reduced by about 25% compared with the traditional method,and the relative error of visual measurement is controlled within 0.2%.