Research On Geometric Dimension Detection Method Of Typical Parts Based On Image Processing

The measurement tools of the traditional part-size detection technology are single,inefficient and low precision,which cannot meet the requirements of rapid and accurate measurement of typical parts size in modern machinery production.The new-type part-size detection system equipment is expensive in manufacturing,maintenance and repair cost.The modern part-size detection technology based on image processing can not only meet the double requirements of high measurement accuracy and high detection efficiency,but also has the advantages of non-contact measurement,which can seriously avoid the secondary damage of typical parts.According to the classification of typical parts,this paper selects the standard spur gear with the broadest application range and the highest frequency as the research object,designs the gear geometric size detection system based on image processing and measures its geometric parameters and single errors.The main research results of this paper are as follows:Firstly,appropriate industrial camera,lens,light source and other hardware equipment were selected according to the gear geometric size and measurement accuracy,a gear geometric size detection platform was built to collect gear images.Secondly,it focuses on image enhancement,ordinary filter noise reduction,threshold segmentation,pixel-level edge detection operators and sub-pixel level edge detection algorithms in image processing technology,and independently designs the gray-level histogram equalization,the hybrid filter combining adaptive median filter and bilateral filter,threshold segmentation based on the maximum inter-class variance method and the adaptive Zernike moment sub-pixel level edge detection algorithm based on Canny operator.Therefore,the gear profile with little noise and clarity can be obtained in this image and because the edge is thicker,it needs to adopt edge thinning.Thirdly,the camera imaging principle and standard camera calibration methods are carefully discussed.The high-precision standard measuring block is selected as the reference for camera calibration,the relationship between actual size and pixel distance is constructed and the calibration coefficient is obtained.The pixel values of the gear center point coordinate,inner circle radius,tooth top circle radius,tooth root circle radius,dividing circle radius and the actual values of tooth number,modulus,tooth pitch,tooth thickness and tooth slot width are obtained by MATLAB design program,which needs to be derived the radius of the inner circle,the radius of the tooth top circle,the radius of the tooth root circle,the radius of the dividing circle of the actual values according to the calibration coefficient.On the basis of determining the tooth pitch and tooth thickness,two actual deviations of tooth pitch deviation and tooth thickness deviation are measured.Finally,the visual interface of the gear geometry dimension detection software system was designed.Through the repeatability accuracy test,a total of ten repeated measurements were taken.The experimental values are compared with the theoretical values measured manually,the geometric parameter errors were maintained within0.02 mm and the single errors were within the allowable deviation range.The standard error of the gear geometry dimension detection system is less than 0.004 mm and the uncertainties are all less than 0.003 mm,proving the system’s accuracy and stability.