Research Of Micro-scratch On-line Detection Methods For Transparent Glass Flat Surface

Due to technical conditions,the production environment and other human factors,the screen of mobile intelligent products appears such as inevitable scratches,chipping,bubbles and other quality problems during its manufacturing,cutting and re-processing.These defects not only affect consumers' selection to the product,also decreasing the optical transparency and consistency of the screen,greatly reducing its mechanical strength and the value of the product.Therefore,picking out tho se screens which not meet the standards from goods and make them rework or scrap through rigorous detecting process is necessary.Low efficiency,big labor intensity,and not being digitized detected results of the traditional manual inspection,can't meet the business requirements of rapid production.To this end,how to detect those defective products from the bulk of transparent glass panels quickly and accurately will be an urgent problem.Except for consulting large numbers of articles about online detecting and similar intelligent detection equipment,this paper researched several key technologies of product surface online detecting based on machine vision by taking micro scratches of transparent panel as an example:Proposing a system architecture us ing industrial computer as control computer and server as image processing unit: industrial computer is responsible for receiving the image processing result data and the feedback signal from sensors,communicating with PLC servo system,releasing image acquisition signals and controlling the position of image acquisition device as well as other executive mechanisms;the server plays its advantages of high-speed data processing,operating the massive image data and sending the results to industrial computer.In this way,industrial computer and the server assume their skilled work thus making better service for the online detecting system.Calculating and selecting the key hardware modules of machine vision detecting system and proposing the belt conveyor to achieve pipelined continued detecting;Putting up a prototype which can be simulated as surface defect detection environment,its detection accuracy is up to 40?m × 50?m when running at the rate of 30 mm / s.Designing a surface scratch detection algorithm based on Halcon,sequentially denoising,thresholding,expanding,feature extracting to acquired images,while designing parameterizes and defining standards for these different algorithms focusing on solving the efficiency and robustness issues of scra tch detecting algorithm.Test results show that the screen surface scratches detection algorithm has high efficiency and partly robustness,satisfying the indicators of efficiency and accuracy involved in the project.Based on the test environment and scratch detecting algorithm,designing and developing a suitable surface defect detection system software,which achieves on-line Real-time monitoring towards the detection data and flexible control to executive mechanisms of surface defect detection device through friendly interface.Experimental results show that the designed transparent panel detection system can basically meet the design expectations,which confirmed the surface detection system's feasibility applied to engineering practice,providing a strong persuasive for the further invest to the project.