Find in Library

Thin Film Transistor Liquid Crystal Display (TFT-LCD) has excellent properties such as lower voltage to start and less occupied space if comparing with traditional Cathode-Ray Tube (CRT). But screen flaw points and display color deviation defects on image display exist in TFT-LCD products. This research proposes a new automated visual inspection method to solve the problems. We first use multivariate Hotelling T2 statistic for integrating coordinates of color models to construct a T2 energy diagram for inspecting defects and controlling patterns in TFT-LCD display images. An Ant Colony based approach that integrates computer vision techniques is developed to detect the flaw point defects. Then, Back Propagation Network (BPN) model is proposed to inspect small deviation defects of the LCD display colors. Experimental results show the proposed system can provide good effects and practicality.