A fuzzy neural network approach for quantitative evaluation of mura in TFT-LCD

摘要

Mura is a typical region defect of TFT-LCD, which appears as low contrast, non-uniform brightness regions, typically larger than a single pixel. It is caused by a variety of physical factors such as non-uniformly distributed liquid crystal material and foreign particles within the liquid crystal. As compared to point defect and line defect, mura is relatively difficult to be identified due to its low contrast and no particular pattern of shape. Though automatic inspection of mura was discussed in many literatures, there is no an inspection method could be used to practical application because the defect models proposed were not consistent with the real ones. Since mura is of strong complexity and vagueness, so it is difficult to establish the accurate mathematical model of mura. Therefore, a fuzzy neural network approach for quantitative evaluation of mura in TFT-LCD is proposed in this paper. Experimental results show that a fuzzy neural network is very useful in solving such complex recognition problems as mura evaluation.