A Defect Inspection System For ITO Circuits Of LCD Based On AOI

Nowadays, the LCD screen has been more widely used with the development ofmodern information display technology, however, the LCD panel production process iscomplex and there are many different kinds of defects, combined with the performanceof today’s inspection equipment, it is difficult to detect the defects automatically in theproduction process of the liquid crystal screen, so the defects are mainly detected bymanual at present. But the efficiency of artificial detection is low and error rate is big,therefore，the automatic inspection of defects generated in the LCD production line isthe key research area at home and abroad, and it is also the mainstream trend of LCD’sdefect detection in the future.In this dissertation，an automatic optical inspection system for indium tin oxidecircuits on LCD glass substrate is researched, it is a comprehensive detection systemwhich needs high requirements for optics, motion control, image processing andsoftware design， and it reflects the intersection of interdisciplinary knowledge, forthis reason, it has a very high research value. The main tasks in this dissertation are asfollows.Firstly, the composition principle of LCD’s circuit inspection system is introduced.The various modules of the system are designed in detail according to the systemrequirements, and the hardware system of LCD’s circuit detection has been fabricated.Secondly, the defect detection and localization algorithms are researched. Theyinclude the acquisition process of standard template image, image filtering andconnected components labeling. Especially, the calculation of anchor points and thebinary image contour tracking are mainly discussed, and the defect detection andlocation is realized.Thirdly, the classification and identification of the defect is studied. After theintroduction of defect characteristics and related theory, it focuses on how to usevariance and connected components analysis for the classification and recognition ofthe defects, and the feasibility of this method has been demonstrated by the experimentresults. Finally, the basic framework of the detection software is explained, then thegraphic user interface of this software is displayed, and the implementation of theunderlying image acquisition and motor control is discussed, in addition, the OpenCVlibrary which is the basis of the image algorithm of this system is introduced.The characteristic of this dissertation includes the following aspects: the motionplatform is driven by linear motor, it has higher stability and positioning accuracy thanthe conventional rotary motor combined with screw rod; the validation andimplementation of the algorithm are all based on OpenCV, compared with thetraditional development mode that validation is based on Matlab but implementation ison other platforms, the efficiency of software development is greatly improved;furthermore, the defect classification method of the ITO circuits based on connectedcomponents analysis is proposed in this dissertation, as far as we know, it is the firstreport in China.