Development of Automatic TFT-LCD Mura Defect Detection

• Main Authors: Chia-Cheng Kuo, 郭家成
• Other Authors: 陳亮嘉
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/7ru4eh

碩士 === 國立臺北科技大學 === 自動化科技研究所 === 94 === An innovative TFT-LCD defect detection algorithm is developed for automatic detection of Mura defects based on Discrete Cosine Transform (DCT) principle for background image reconstruction. Efficient and accurate surface defect detection on FPD panels has never been so important in achieving the high yield rate of FPD manufacturing. All kinds of FPD manufacturers need to inspect products such as screen panels, control PCBs and final assembly modules. Front-of-screen (FOS) quality performed by a human visual inspection is susceptible to unacceptable manufacturing costs and uncertain product delivery time. Therefore, automatic inspection of FOS quality is highly essential to achieve effective defect detection for optimizing operation efficiency and product quality. One of the visually most difficult recognizing problems in LCD panel inspection is to deal with clearly distinguishing specific regions of low contrast and non-uniform brightness called mura. A mura defect in general processes a non-uniform brightness region that slightly differs from the background by down to unit signal level, where it is detectable only when its size is larger than a specific size. Detecting blob-Mura defects in a LCD panel can be difficult due to non-uniform brightness background and slightly different brightness levels between the defect region and the background. To resolve this issue, a DCT-based background reconstruction algorithm is developed to establish the background image without mixing with the detected object- Mura defects. Based on DCT principle, we present a new segmentation method for detecting area-mura. Through some experimental tests on natural Mura defects, it was verified that the proposed algorithm has a superior capability for detecting blob-mura defects in its detection accuracy and speed.