Research On Key Methods Of Visual Precision Measurement Of Shaft Parts

With the continuous advancement of industrial automation,automotive,medical,consumer and other industries have put forward new requirements for product quality.Shaft parts are the core components of products in various industries.The machining accuracy directly affects the operating status and service life of automated machinery and equipment.Especially in precision instruments,the dimensional accuracy of shafts often reaches micrometers.Because China still has obvious deficiencies in the precision measurement of shaft shafts,it has aggravated the dependence of Chinese companies on foreign technology,especially in terms of corner point and R-angle location methods,which is a far difference from foreign countries.Therefore,this article deeply studies corners and R-angle location technology is of great significance.Aiming at the problem of interference of corner detection in vision measurement on the surfaces of shaft parts such as burrs and oil stains,a corner sub-pixel location method based on curvature and gray combination is proposed.This method performs morphological and bilateral filtering preprocessing on the Region of Interest(ROI)of the image to eliminate fixtures such as burrs and partial oil stains;it detects candidate corners based on the curvature characteristics and utilizes multi-scales of curvature angles at the corners.Invariance: pre-screen false-corners,use the gray information in the circular window centered on the corners to further exclude the false-corners,and achieve coarse-corner location;according to the connection line between the coarse location corners and the end points of the area,the original The edge points of the image are filtered,and the filtered edge points are fitted with least square straight lines to achieve precise location of the corners.Experiments show that this method effectively overcomes the interference problem of the attachment on the surface of the shaft parts,and the repeatability of the corner location algorithm reaches 0.005 mm and the accuracy reaches 0.004 mm.Aiming at the interference of R-angle detection in visual measurement and the instability of short circular arc fitting for contour attachments,an arc fitting method based on Snake model and iterative Polarity Transformation Regression(PTR)was proposed.First,ROI is preprocessed;the edge extraction is performed using the Canny operator to achieve the initial location of the edge;then the edge is optimized by the Snake model;the sub pixel level edge points are acquired by the Zernike orthogonal moment;and finally,Finally,an iterative PTR algorithm is used to achieve accurate location of the R-angle center.Experiments show that this method effectively overcomes the problem of interference of R-angle detection by attachments on the shaft surface and improves the repeatability and accuracy of the R-angle location algorithm.The location repeatability is 0.005 mm and the accuracy is 0.0034 mm.The fitting repeatability is 0.002 mm and the accuracy is 0.006 mm.In order to verify the comprehensive performance of the measurement system,a visual inspection hardware platform for shaft parts was built using line-array cameras,telecentric lenses,and parallel backlights.Multiple detection items for multiple motor shafts were tested and analyzed,from static and pick-and-place.Get the image of the object under the condition,calculate the extremum of the static image result to calculate the repeatability of the system measurement,and calculate the accuracy of the system measurement by calculating the maximum error between the result of capturing and releasing the image and the real value.Finally,compare the experimental results and analyze its synthesis.Measurement repeatability,accuracy,and error factors.The experiment shows that the repeatability of the plane size measurement of the shaft parts measurement system reaches 0.007 mm and the accuracy reaches 0.05 mm,which meets the customer's detection requirements.