Research On TFT-LCD Micro Defect Detection Method Based On Image Processing Technology

Compared with traditional CRT(Cathode Ray Tube) displays, TFT-LCD (Thin Film Transistor-Liquid Crystal Display) characteristic such as high resolution,high lightness,no geometric deformation,small volume light weight and low power, as the new generation flat display , TFT-LCD have powerful competition advantage. Price is one of the major factor of influence to LCD market. For reducing the LCD production costs and increasing the LCD yield, we must develop real-time and high-efficient defect detection methods. The manufacturing process of TFT-LCD is comprised of about 100 working procedures. Though most of them are performed in clean rooms, TFT-LCD panels would exhibit some defects inescapability. Most conventional TFT-LCD defect detection methods are effective for macro defects, but not for micro defects,which is far smaller than a display pixel and can not seen by human eyes.In this paper, for detect micro defects which are not visible by human eyes , we propose a TFT-LCD micro defect detection method based on local template matching. For implementing image registration in sub-pixel level, a image registration method based on local template matching is proposed. On the base of this method, we propose a TFT-LCD micro defect detection based on the image substraction method and sub-pixel image registration. With this method, we can detect TFT-LCD micro defects automatically.Applying this method in TFT-LCD Array Area Detection and Peripheral Area Detection, the results show that the rates of both miss and false detection in array area is lower than 3%, and in peripheral area is lower than 5%. This method can meet the detection requirement.