Detection of Spot-Type Defects on Liquid Crystal Display Modules

In the manufacturing process of a LCM(Liquid Crystal Display Module), many spot-type defects can be occurred on the surface of LCM due to various physical factors. The existence and pattern of such defects are very important in determining whether the LCM is normal or not. To enhance the accuracy and reproducibility of LCD inspection, this paper introduces an automated inspection method using a computer vision technique. The LCM defects are classified into macro-defects and micro-defects. One is detected by using a macro-view area camera and the other by using six micro-view line cameras. An adaptive multilevel thresholding method using statistical characteristics of local block is proposed for a macro-view image while the detection method for a micro-view images composed of R, G, B sub-cells involves a pattern elimination technique using the pixel difference and adaptive multilevel thresholding. The proposed inspection system is tested using many real LCMs having different defects, and the resulting performance confirms the effectiveness of the proposed algorithm. Introduction The recent rapid development of FPD (Flat Panel Display) devices is demanding manufacturing processes with a high production yield and low production costs. In particular, TFT-LCD panels are expanding faster than other FPD devices in terms of size and density, thus TFT-LCD panel suppliers need a higher accuracy and speed for LCD production. But, current final inspection tests rely on human inspectors. Thus, the inspection results can be varied with the inspectors, moreover the same panel can appear quite different at different times, even with the same inspector. Therefore, an automated LCM inspection system for detecting various types of defects in LCD panels is needed to enhance the inspection accuracy and reduce the production costs [1]. In this paper, an automated inspection algorithm is presented for detecting spot-type defect in TFT-LCD panel using a computer vision system [2]. However, certain considerations need to be addressed for detecting spot-type defects. First, in the case of high-resolution image, the cells have an obvious R, G, B sub-cells pattern, which needs to be eliminated before proceeding to next stage. As such, a difference image is used based on the pattern period that is estimated by the MSE (Mean Square Error) of the block difference. Second, the background brightness of an LCD panel is non-uniform, even when a constant signal is driven to the whole panel, because of the backlight position and imperfection in diffusion plate. Therefore, the background and defect regions in an acquired LCD image both have non-uniform brightness regions. Thus, the inspection specifications for an automated inspecting system are very critical. A spot-type defect, which differs from the background region (normal region) merely by 1 signal level, is only visible if it is bigger than a specific size. Consequently, the current study employed an adaptive multilevel threshold method [3] that uses the statistical characteristics of the local area for adaptive segmentation [2] of spot-type defects. Key Engineering Materials Online: 2004-08-15 ISSN: 1662-9795, Vols. 270-273, pp 808-813 doi:10.4028/www.scientific.net/KEM.270-273.808 © 2004 Trans Tech Publications Ltd, Switzerland All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications Ltd, www.scientific.net. (Semanticscholar.org-13/03/20,17:37:59) Title of Publication (to be inserted by the publisher) The remainder of this paper is organized as follows. First, the spot-type defects that appear in TFT-LCD panel are described. Second, a detection scheme using adaptive multilevel thresholding is outlined for low-resolution images, followed by a detection scheme for very small defects using high-resolution images acquired by a line-scan camera. Thereafter, some experimental results are presented. Finally, the paper is summarized along with some concluding remarks. Spot-Type Defects in LCD Spot-type defects appear as blemish in an LCD panel, have diverse sizes and shapes, and are caused by variable physical factors. Generally, when a screen is driven with a constant signal level, such blemishes are visible with the naked eye. However, there are also very small defects, called point defects, which are invisible in a low-resolution image, yet, detectable in a high-resolution image using a line-scan camera. Therefore, spot-type defects can be divided into two categories: blob defects, which are relatively large and can be observed in a low-resolution of a spatial domain image, and point defects, which are relatively small that can be observed in a high-resolution image. Blob Defects. Blob defects cover a region of more than 1 pixel and have a different gray level from the surrounding normal region. Blob defects that contrast strongly with the normal region are easy to detect, yet very weak blob defects, just 1 level above or below the normal region, are very difficult to detect automatically. The main reason for this difficulty is related to the quality of the acquired image. If the edges are noisy, it is extremely difficult to distinguish noise from a defect and the brightness of LCD surface is not uniform even if a constant signal is driven into the panel. Fig.1 (a)(b) shows some example image including blob defects. (a) (b) (c) (d) Fig. 1. Blob defects and point defects. (a) Real blob defect displayed in black circle. (b) Artificial blob defects with varying size and contrast. (c) High-resolution image of LCD. (d) A point defect. Point Defects. Point defects are caused by an inferior gate, or disassembled color filter, etc. When a high-resolution image is acquired with a line scan camera, it can be seen that an LCD module is composed of many cells and each cell is composed of three sub-cells, representing red, green, and blue. Generally, point defects are too small to take an accurate detection with naked eye. Therefore the human inspectors use a magnifying glass, such as a Lupe. As such, the proposed automated LCD inspection system uses the high-resolution line scan camera. Fig. 1 (c)(d) shows a high-resolution LCD surface image using a line scan camera and an example point defect. Detection of Blob Defects. The detection of blob defects involves many difficulties. Among these difficulties, the most severe one is the non-uniform brightness of the LCD surface. To solve this problem, this study employs inter and intra image enhancement and an adaptive multilevel-threshold technique based on local blocks. Fig. 2 (a) shows the flowchart of the proposed method. In the acquired image, the average brightness of the LCD can differ dramatically even with the same model. Therefore a preprocessing stage is applied first, involving inter and intra image enhancement. After the preprocessing stage, an Key Engineering Materials Vols. 270-273 809