Research On Detection Method Of Processing Accuracy Of Bobbin Based On Machine Vision

As the core part of the yarn feeder of computerized flat knitting machine,the coaxiality of the yarn feeder has a great influence on its performance.Today,with the rapid development of manufacturing technology,the requirements of detection technology are constantly improving.The traditional manual contact part size detection method is slow,low precision and poor flexibility,which can not meet the current detection needs.In the trend of industrial intelligent development,it is of great significance to realize the rapid and accurate detection of the processing accuracy of the bobbin for the intelligent manufacturing of the bobbin.Therefore,according to the structure characteristics and processing inspection technology requirements of the bobbin,based on the machine vision inspection technology,this paper proposes an efficient and accurate inspection method for the processing accuracy of the bobbin,and designs a set of processing accuracy inspection system,which not only improves the inspection speed and accuracy,but also ensures the reliability.To this end,this paper mainly completed the following aspects of research work:First of all,the visual inspection platform of the bobbin is designed and built to complete the image acquisition.Based on the analysis of the characteristics and inspection requirements of the bobbin workpiece,combined with the machine vision inspection technology,the inspection scheme is determined,and the hardware of the inspection system is optimized.Then,the experiment platform of camera distortion correction is established.Secondly,an improved Canny edge detection algorithm is proposed to preprocess and detect the edge of the collected image.In order to reduce the amount of image data,the acquired image is grayed,and the median filter method is used to improve the effect of edge preservation.Due to the long tube depth of the feeding tube,the exposure of the pneumatic claw at the bottom is insufficient,so the contrast of the feeding tube edge and the pneumatic claw is enhanced in turn,and then the enhanced image is combined to achieve the effect of enhancing the image contrast.Then the traditional Canny edge detection algorithm is improved from three aspects of filtering method,gradient direction calculation and threshold setting.By comparing with Roberts,Sobel,canny and other traditional edge detection algorithms,it is found that the improved Canny edge detection algorithm can achieve better edge detection.Then,the RANSAC & least square method is fused to fit the detected image edge.In order to reduce the measurement error,the detected edge features are fitted.After analyzing the shortcomings of the traditional least square method in the edge fitting,the edge fitting algorithm of RANSAC & least square method is proposed.The central response surface experiment is designed to obtain the linear regression model between the fitting parameters and the fitting results,and the optimal combination of fitting parameters is obtained.The fitting results show that RANSAC & least square method can better remove the influence of external points on the fitting,and the fitting edge is closer to the actual edge of the image,which can better reduce the error of visual detection.Finally,the software system is designed and developed,and the experimental verification is completed.From the perspective of application,MFC library and opencv open source vision library are used to develop a machine vision based yarn feeding tube processing accuracy detection system,which realizes camera calibration,image processing,edge fitting,processing accuracy detection and other functions,and is convenient to use.Then the system is used to measure the size of five groups of bobbins.Compared with the actual measurement results,the absolute error of the detection accuracy of the inner and outer diameter of bobbins is between0.04-0.09 mm,and the detection result of coaxiality is also within the required range,which fully meets the requirements of high efficiency and accuracy of the processing accuracy detection of bobbins.The causes of the errors are analyzed and classified,and the improvement measures are put forward.