Inspection of thin film transistor (TFT)-patterned glass is one of the important processes for TFT-LCD panel manufacturing. Minute pattern defects and particles coexist on the glass and a few defects can cause serious quality problems. If there was a way to classify the defects by its potential fatality, it would be useful to control the product quality and loss of inspection time. This article introduces a rule-based approach of zone-based inspection and defect classification. For the zone-based processing, a defect classifier and its decision boundary were also proposed to differentiate non-fatal defects. An experimental result showed that the defect classifier was useful to select non-fatal defects and over 70% of the trivial defects could be selected with little risk of quality problems. This article also presents an application result to utilize the method in a plant. Since prejudged trivial defects have little influence on panel quality, they could be eliminated without interfering with the factor effecting overall final quality. This enables to save post-inspection time. The application results are sufficiently promising to save over 30% in inspection time and, furthermore, only 0.02% of fatal defects are not detectable. The proposed method also brings about an additional return of yield improvement because operator-error is found to be decreased as the defect candidates decrease because of the application of the proposed method.
LCD(Liquid Crystal Display) became one of the most popular display devices in these days. The TFT(Thin Film Transistor) substrate is the key part of active matrix LCD. TFT is an electrical device to activate a displaying cell. To display an image precisely, several millions of identical transistors are patterned on a wide glass panel. Since a minute damage on the pattern can causes a serious defect to display, it is important to inspect the pattern precisely. Taking the advantage of the fact that the pattern of good cell should be identical to that of adjacent cells, it would be a convenient way to compare a cell with its neighbor cells to find a defect. In practical applications, if the period of repetition could be represented as an integer number of digitized image pixel, it would be possible to find a damaged pixel readily. However, the period of pattern depends on the product size and cannot be determined as an integer always. In this paper, so called, pseudo-matching magnification algorithm has been introduced to solve the problem. A digital image was magnified and period of pattern can be determined as an integer from the processed image. It has been shown that the defects could be enhanced after the preprocessing of digital image. As a result, a TFT-pattern inspection system has been developed and it has been shown the proposed method is compatible for the inspection of repeated pattern.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.