Research On Mura Detection In TFT-LCD

Mura is a familiar kind of visual defect in TFT-LCD, which is understood as defect with low contrast, non-uniform brightness regions, vague contour, and typically larger than a single pixel. It imparts an unpleasant sensation to viewers. Meanwhile, Mura is the most complicated defect so it is very hard to be detected. Currently, most of final inspection has been done by experienced human inspectors with limit samples of Mura in TFT-LCD industry. It has some drawbacks that each observer tends to have subjective decision inevitably due to the fallibility of human perception. Furthermore, it is inefficient.Recently, researchers expect to detect it using machine vision instead of human eyes, but how to detect it correctly is always one of difficulties accepted by TFT-LCD industry. Therefore, the paper analyzes the character and rules of Mura and proposes a method of obtaining the information of Mura based on B-spline surface fitting. During the research, an inspection system based on the method, including the corresponding hardware and software, is designed and established. Performance of the proposed method is evaluated on many real TFT-LCD panel samples. The experimental result shows it is an effective method to detect Mura automatically with high detection rate.The achievement of the paper can offer an effective and valuable approach to Mura detection realizing machine instead of human. Meanwhile, it is also a valuable way for defect detection of other FPD.