Research On Edge Detection And Positioning Technology Of Convex Arc Grinding Wheel Based On Machine Vision

At present,in terms of the research of grinding wheel detection,scholars mainly focus on the detection of grinding wheel’s topography and landform.It is difficult to directly obtain the size information and deviation of grinding wheel contour because of the large amounts of data and the low efficiency of detection.What’s more,most detection methods have the problem of installation error when reinstalling the grinding wheel.However,machine vision detection method does not have these problems.In this thesis,the ceramic bonded grinding wheel is taken as the research object,and the convex arc profile is taken as the shaping target.A machine vision based grinding wheel profile detection system is built.The system can measure the shape and size of the grinding wheel and analyze the shaping deviation without dismantling the grinding wheel,and compensate the shaping process to improve the shaping quality.Firstly,a grinding wheel contour detection system based on machine vision was built,which consists of industrial camera,lens,illumination system and detection mechanism.Based on the grinding copy method,the image information of the contour of the grinding wheel was obtained through the machine vision detection system.Secondly,an improved edge detector based on parabolic interpolation was proposed.The filtering and gradient calculation process in Canny algorithm was improved.Parabolic interpolation was used to calculate the sub-pixel edge position to achieve accurate positioning,which provides the basis for dimension measurement and matching positioning.Compared with the traditional method,the maximum error of the edge detector for linear feature detection is reduced from 0.124 pixels to 0.066 pixels.In this system,the roundness of the arc feature detection is 3μm.The obtained edge points are smoother and more precise.Thirdly,the positioning method of the convex arc wheel image was designed,which is based on the template matching algorithm and the edge feature detected by the edge detector in this thesis.By this method,the pose transformation matrix of the contour to be measured and the design contour were found,and the dimensional deviation of the transformed contour and the designed contour was calculated.Calculate the dimensional deviation of the transformed contour and the designed contour.Fourthly,a software for convex arc diamond grinding wheel detection was designed.The software includes image acquisition module,image preprocessing module,camera calibration module and matching positioning module.The shaping experiment of convex arc ceramic bond diamond grinding wheel was carried out.The experiment proves that the measurement error rate after calibration of the detection system is within 0.022%.The maximum deviation value obtained by measuring the grinding wheel after rough shaping is 126.19μm.The error rate of fitting arc radius is 0.782%,and the roundness is 18.11μm.The maximum deviation value obtained by measuring the grinding wheel after compensation and shaping is 36.96μm.The error rate of the fitting arc radius is 0.036%.The roundness is 14.06μm,and the shaping quality is improved.