Research And Realization Of Bearing Seal Ring Size Measurement System Based On Machine Vision

Rolling bearings play an important role in the machinery industry.The rubber sealing rings play a sealing role on the grease-lubricated rolling bearings,which not only prevents the internal grease from flowing out,but also prevents external particles from entering the bearings,which is of great significance to prolong the actual service life of the bearings.The tolerance range of the outer diameter of the bearing seal ring is 0.12 mm,and the tolerance range of the inner diameter is 0.16 mm.The inner and outer circular lips of the bearing seal rings are made of elastic rubber material,which cannot be measured with traditional vernier calipers.Enterprises generally use hand-cranked optical image measuring instruments to measure the inner and outer circle diameters of the seal rings.This measurement method has low measurement accuracy and its efficiency is slow.In order to improve the measurement accuracy and efficiency,this paper proposes a sub-pixel measurement method for the seal rings based on machine vision,builds a test bench for the seal ring measurement,and develops the measurement software.The main work content is as follows:According to the size range of the sealing rings and the measurement accuracy requirements,choose suitable industrial camera,industrial lenses,and light sources,and build a size measurement test bench with greater rigidity.In order to better describe the relationship between the coordinate points in the three-dimensional world and the corresponding pixels in the image,a linear model of camera imaging is established.Taking into account the influence of lens distortion,distortion parameters are introduced,and the linear imaging model is corrected to a nonlinear imaging model according to the Taylor expansion of unary function.By deriving the mathematical model of camera calibration in detail,it is concluded that the internal parameters of the camera and the external parameter matrix between the image and the camera can describe the mapping relationship between the target and the two-dimensional image in the three-dimensional world,and can correct the distortion of the image.Use Zhang Zhengyou calibration method in Matlab software to obtain the camera’s internal and external parameters matrix.The Harris corner detection method is used to detect the corners of the reference pose image of the calibration version after the distortion correction,and obtain the equivalent pixels to prepare for the subsequent size measurement.Perform simulation experiments in Matlab,using mean filtering,median filtering,Gaussian filtering,and bilateral filtering algorithms to filter and denoise grayscale images containing salt and pepper noise and Gaussian noise,and use median filtering according to the filtering effect and filtering time.The algorithm performs noise reduction processing on the gray image of the seal rings.According to the threshold segmentation effect of the histogram double peak method,iterative threshold segmentation method,and the maximum between-class variance method on gray-scale images,the maximum between-class variance method is determined as the threshold segmentation method for the image in this article.Aiming at the problem of the edge detection of the target in the image,three basic models of the image edge are established,the mathematical principle of the existing pixel-level edge detection algorithm is studied,the GUI interface of the image edge detection is written using Matlab,and the Roberts algorithm,the Prewitt algorithm Sobel algorithm,Lo G algorithm and Canny algorithm are compared on the edge detection effect of gray image,it is found that Canny algorithm has a stronger detection ability for fine edges.The mathematical principles of the sub-pixel edge detection method based on the moment method,fitting method and interpolation method are analyzed,and an improved sub-pixel edge detection method is proposed according to the characteristics of the ring shape of the sealing ring.The core part of the method includes gray-scale interpolation,Bool area operation,Canny edge detection and least square fitting steps.First of all,the measurement results of the inner and outer circle diameters of the sealing rings by the three-coordinate measuring instrument is the true value.Furthermore,the measurement results of the inner and outer circle diameters of the same sealing ring by the hand-cranked optical image measuring instrument and the sub-pixel method in this article are the measured values.In addition,the absolute error of the method and the time-consuming measurement process of the latter two measurements are counted.The absolute measurement error of the sub-pixel method in this paper is less than 0.058 mm,and the average time to measure the two diameters of a sealing ring is 2.91 s.The measurement accuracy and measurement efficiency are better than the measurement method of the hand-cranked optical image measuring instrument,which proves that the sub-pixel measurement method proposed in this paper has practical significance and application value.