Research On Key Techniques Of 3D Vision Guidance For Shoe Glue Spraying Based On Point Cloud

Automation of shoe mold spraying is an urgent need for shoe manufacturers.The rapid development of science and technology in recent years has provided technical support for shoe companies to realize automation and intelligence.Compared with the image information,the three-dimensional information can more comprehensively and truly reflect the shape of the shoe mold.In combination with visual guidance technology,it is possible to adjust the spraying method according to the posture of the shoe while identifying and positioning the shoe mold.This article takes shoe molds as the research object,and focuses on two key techniques in the spraying of shoe molds—the method of extracting the center of line structure light stripes and the method of extracting the 3D trajectory of point clouds.The specific work includes the following aspects:(1)The center of light strip extraction in the image is obtained by using the light strip center extraction algorithm of double-frequency curve fitting.First,using the threshold method twice to automatically segment the light fringe area in the image,the gray center of gravity method is used to obtain the approximate light stripe center,and the local normal of the stripe is obtained by fitting the approximate light fringe center.Then,using dual-frequency curve to fit the pixel data in the normal direction,and finally,using golden section method to search the maximum value of the light stripe region of the fitted curve,and the sub-pixel coordinates of the light stripe center are obtained.This method can eliminate the influence of some high-frequency noise in the image.Compared with the Steger method,the average value of the center of the light strip is less than 0.1pixel,and the speed is increased by about 25 times.(2)A variable intercept contour migration algorithm based on point cloud projection is used to extract the three-dimensional spray track of shoe molds.According to the characteristics of the point cloud obtained by line structured light scanning,pass-through filtering,statistical filtering,and grid filtering are used to preprocess the point cloud of the shoe model.The point cloud contour of the flat shoe mold was obtained by using the data dimensionality reduction method of point cloud projection and principal component analysis,and the projection plane spraying trajectory is obtained through the variable intercept contour offset algorithm and the B-spline curve fitting method.Finally,the 3D spraying trajectory of the shoe model is obtained through the inverse mapping and point cloud normal estimation algorithm.The three-dimensional trajectory extraction algorithm can be used not only to obtain the trajectories of shoe mold spraying,but also to extract trajectories of complex curved products.(3)A line-structured light-scanning glue vision guidance system for shoe molds was set up.A multi-ball three-dimensional calibration target suitable for three-dimensional hand-eye calibration is designed,and the calibration of the vision guidance system for spray injection of shoe molds is completed.Under the Windows7 system platform,the MFC framework was used to develop the glue spraying system software.The three-dimensional vision guidance system can not only scan and extract three-dimensional glue spray trajectories of shoes with different specifications in real time,but also implement three-dimensional trajectory guidance tasks of different curved surface parts by loading parameter files.From the operating results of the built 3D vision guidance system,the line structure light stripe center extraction algorithm and trajectory feature extraction algorithm studied in this paper have high operating efficiency and certain universality,and can be widely used in shoe mold spraying and Other 3D vision guidance.