Research On The Measurement Method Of Large-size Plates Based On Machine Vision

Large-size sheet metal is one of the indispensable industrial raw materials in production and construction.It is widely used in the manufacture and assembly of maj or equipment such as aircraft,ships and large-scale mechanical equipment.Accurate measurement of the sheet metal is a key step in ensuring the accuracy of equipment manufacturing.With the rapid development of machine vision technology,vision-based measurement systems have gradually become an important measurement method in the industrial field due to their advantages of non-contact,high precision and high speed.Aiming at the residual material management and blanking and nesting of large size sheet,a method of measuring sheet plane size based on machine vision technology is proposed.Firstly,a measurement system consisting of a three-dimensional robotic arm and a line scan camera is designed.The robotic arm can move on three dimensions and collect complete and clear large-size plate images in the measurement area through the motion control system.Secondly,a calibration method for line scan cameras is proposed.Based on the analysis of the imaging principle of line scan camera,a specific field of view is selected as the calibration area in the moving direction of line scan camera.A certain number of calibration plate images are obtained in the calibration area,and the likelihood function is constructed according to the feature points on the image.Combined with levenberg-marquardt algorithm,the optimal solution is obtained iteratively,and the internal and external parameters of the line scan camera are calculated.The effects of the changes of camera scanning speed and the number of calibration plate images on camera calibration are studied.Then,different edge detection algorithms and sub-pixel edge calculation models are analyzed.Canny edge detection and grayscale moment algorithm are used to locate the sub-pixel edge of image.The edge of sheet is divided into a certain number of geometric primitives by polygon fitting method after several iterations.The divided edge is fitted with the least square fit algorithm based on the weight function and the large size plate is accurately measured according to the fitting result.Finally,an error compensation model of the measurement system based on the plane error field is established by analyzing the causes of the errors of the measurement system.A grid standard plate is designed and the error distribution on the grid line is obtained by using the measurement results of the standard plate.According to the distribution,the error curve is established,and the trig function theorem is used to project the curve into the image pixel coordinate system.The error field of the whole measurement area is reconstructed by linear interpolation,and the measurement results are compensated and corrected.On the basis of these theories and technologies,the sheet metal is measured in the measuring area of 1.2m x 2.6m.The experiment shows that the precision of the measuring system can reach lmm/m,which meets the requirements of accuracy and speed of field measurement of large-size sheet metal industry,and has a high application value.