Design And Research On Vision System Of A Double Dispensing Machine

With the continuous development of China's high-tech electronic information industry,the demand for automatic dispensers is also growing.In order to improve the production efficiency,the traditional single valve contact dispenser is being converted to double valve non-contact dispenser.As the double dispensing technology is still in the stage of research and development and improvement,there is less research on the vision system of this kind of double dispensing in China.In this paper,the vision system of the double dispensing is studied,The main contents of this paper are as follows:According to the requirements of dispensing process,the hardware of camera,lens and light source is selected.The hand-eye calibration scheme is determined by analyzing the characteristics of dispensing machine.The traditional nine-point calibration board is replaced by the corresponding dispensing spraying nine glue points on the calibration board.At the same time,based on the calibration matrix obtained by the nine-point calibration method,a distance measurement method is proposed to measure the distance between the main dispensing valve nozzle and the sub dispensing valve nozzle.Using machine vision method to complete the distance measurement of double injection valve,the process of manually measuring valve distance is saved,and the degree of machine automation is improved.Taking the annular material as the main research object,the image processing algorithm of the image collected by the camera is studied.The gray stretching method is used to preprocess the acquired image.In image recognition and location,based on image cross-correlation algorithm,non-maximum suppression algorithm and Gaussian-pyramid,an automatic ROI recognition scheme is determined,which replaces the existing manual method of ROI division.The implementation process of contour template matching algorithm is analyzed,the similarity measure method is determined,and the rotation invariance and pyramid search strategy are used to accurately locate the object to be measured.Find out the specific position of the material in the image.According to the dispensing process,the selection of theoretical dispensing position is simulated by dispensing experiment,and the best dispensing position selection method based on edge single dispensing is proposed.Based on the minimum circumscribed rectangle method and bilinear interpolation method,the dispensing position of square and oval materials was determined.The visual guidance experiment of double dispensing was completed.After dispensing,Using blob analysis method to identify and detect glue coverage area.Otsu threshold segmentation method is used to segment the image and removing interference area by morphological processing.The feature quantity of the interested connected domain is calculated,and the shape feature of the connected domain is calculated to detect whether the dispensing result is up to the standard.The traditional image subtraction is improved by image calibration,low-pass filtering and contour weighted suppression,and the improved image subtraction.For the materials that are seriously oxidized or corroded and cannot be identified by blob analysis method,the improved image subtraction method is used to extract the connected domain,and the blob analysis method is used to analyze the feature quantity.The calculation of glue coverage area in image detection is completed.The related research and experiment of hand eye calibration,double valve ranging and image processing algorithm are completed by Matlab 2018 software,Halcon12 software and Visual Studio 2015 platform.In the experiment,the normal productivity of defect free ring material is 99.82%,and the normal productivity of defect ring material is 92.76%.