Study And Application Of On-line Detect System Based On TFT-LCD Module Defect

In this thesis, the author mainly studies the on-line defect detection technology of surface with the uniform and homogeneous texture, its main application is TFT module on the thin film transistor-liquid crystal display (TFT-LCD) with homogeneous texture image. At present TFT-LCD defect detection is mostly the off-line mode, and needs a lot of manpower, so its efficiency is very low and also will create the deviation in the process of detection due to the subjective judgment. Therefore, the author in this thesis has introduced the vision system for the surface defect detection of TFT-LCD module by means of the on-line detection mode to enhances the precision and the speed of defect on-line detection in TFT-LCD module.The author combines the research and development of the enterprise for on-line detection system, studied this system and the more details are as follows:1. The author has studied the structure ,characteristics and display principles of TFT-LCD module as well as its module defect production, detection and distinction, and analyzed the current situation and prospects about TFT-LCD module R&D, its production, detection and applications in different domestic and foreign manufacturers.2. The author has studied the influence of the system hardware on the precision and the speed of TFT-LCD module defect on-line detection by analyzing the hardware of the on-line detection system in a Electro-optic Company (including the marble platform, optical system, man-machine interface and so on).3. Research on the AOI system's work, including recipe setting of the parameters , defect scanning , and detection&analysis of the defect. The recipe parameters is mainly refers to the image data, detecting zone, the optical device and the detection sensitivity. After scanning the defect, TDI will take photoes of defect according to the recipe setted before.4. As the optimization method of ICA solution, PSO(Particle Swarm Optimization) has been studied, and the constraint is put into the solution process to make the response value identical after the photos are filtered in the normal zone. By means of the quality management method, according to the thresholds of the response value, the defects will be determined. 5. The results of the on-line detection are analyzed and discussed, the reason and the characteristics of the defects are also induced and summarized. It provides the basis for the defect detection and downstream diagnosis of the TFT-LCD module, and thus enhances its production rate.The research in the thesis is summarized finally, and the future research direction is pointed.