Research And Development Of Automatic Dispensing Control System Based On Vision Positioning

In recent years, with the rapid development of the chip industry at home and abroad, great changes have taken place in mobile phones, tablet computers and other electronic industries. Major mobile phone manufacturers to design products to be thinner and lighter, therefore, the process of industry put forward higher requirements. Traditional labor-intensive production mode have been unable to meet the demand of high precision processing technology, more and more parts manufacturers to develop or introduce automation equipment and machine vision system plays a very important role in the production of automation.In this paper, the research and development of camera automatic dispensing machine, is adding machine vision system based on the existing semi-automatic machine, in order to change the cumbersome manual operation of semi-automatic machines and satisfy the needs of high precision dispensing tasks. The core function of the dispensing control system is realized as the visual orientation and motion control, the bottom motion control is realized by using the open CNC system software. and mainly has completed three tasks in visual orientation. Firstly, the camera calibration algorithm is used to obtain the transfer matrix between the image coordinate and the world coordinate. Secondly, write image preprocessing algorithm, followed by use the Gauss Filter to remove noise, Pyramid Sampling to compress image, Sobel Operator to extract image edge feature. Finally, the template matching algorithm based on edge feature is written. The Generalized Hough Transform is used to locate the template preliminarily, then design filtering algorithm to get edge points set, at last precise localization by using the algorithm of Random Sample Consensus.Combined with the dispensing machine actual mechanical structure and manufacturing process, this paper mainly solves two key points influencing the dispensing accuracy. First, fit the rotation trajectory of the rotation axis and design compensation algorithm, this is because the actual measured rotation trajectory is the egg curve, if it simplified as a circle model will affect the dispensing path planning. Second, confirm the location of the needle glue, in order to prevent glue from blocking the needle in actual production, every four hours must to replace the needle, therefore, it is needed to design a fast and accurate needle vision positioning scheme.In the end designed a simulation approach in order to verify that the software algorithm is correct and meet the demand of precision, this plays an important role for debugging. Actual batch processing tests prove that the cameras automatic dispensing system meet the production requirements in precision, efficiency and stability and so on.