High‐speed TFT LCD defect‐detection system with genetic algorithm

Lin, Chern‐Sheng; Liao, Yo‐Chang; Lay, Yun-Long; Lee, Kun‐Chen; Yeh, Mau‐Shiun

doi:10.1108/01445150810849037

Cited by 17 publications

(7 citation statements)

References 9 publications

(15 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To perform localization Kim et al [5] used adaptive multilevel defect detection and probability density estimation for TFT-LCD inspection. Lin et al [6] presented an image processing method for defect detection in TFT-LCD images and used genetic algorithm (GA) for adjusting heuristics automatically. Ngo et al [7] also presented an automatic detection method for MURA by accurate reconstruction of the background by training separately on the background but using test set images of MURA.…”

Section: Literature Survey For Mura Defect Inspectionmentioning

confidence: 99%

Deep Learning based MURA Defect Detection

Singh¹,

Kumar²,

Sultania³

et al. 2019

EAI Endorsed Transactions on Cloud Systems

View full text Add to dashboard Cite

INTRODUCTION: MURA defects in LED/LCD panels are one of the most challenging defects for Automatic Defect Classification and Localization (ADC) due to their extremely low contrast when compared with the background. Manual detection is subjective, error prone, very tedious and time consuming. Even when the type of MURA defects can be ascertained manually, the exact bounding box for defect is hard to determine. Various heuristic based image processing techniques have been applied giving sub-optimal accuracy over generic datasets. OBJECTIVES: The primary objective of this paper is to check whether the state of the art DL (Deep Learning) network for general object classification and localization (MSCOCO PASCAL VOC etc.) can be applied successfully for MURA Defect Classification and Localization. METHODS: In this paper we present a single DL pipeline for classification and localization which for the first time is applied for MURA defects. Naive DL network-Single Shot multi-box Detector (SSD, pre-trained on ImageNet) was not sufficient to give a good F1 score because of the nature of the defect. Accuracy improved a little after applying various DL specific optimization methods such as loss function optimization, network optimization etc. Utilizing the knowledge from MURA domain for data augmentation, like filtering based on image capture wavelength etc. improved the results significantly. RESULTS: Using optimization techniques that are from both DL domain as well as specific to MURA domain, we show improvement in the accuracy of the base DL pipeline from ~30% to ~80%. Minimum heuristics were used to define the pipeline so that it can easily adapt to any new MURA dataset. The paper shows the importance of domain specific preprocessing steps for the designed network in case of MURA defects. CONCLUSION: Using DL, MURA classification and localization had not been tried before. For the first time we demonstrated results for both classification and localization of MURA defects using state-of-the-art DL network with F1~80%. We also conclude that state-of-the-art network for general object detection can be reused with the help of Transfer Learning (TL) concept and fine-tuned with MURA domain specific optimizations mentioned in paper for optimal performances in MURA domain.

show abstract

Section: Literature Survey For Mura Defect Inspectionmentioning

confidence: 99%

Deep Learning based MURA Defect Detection

Singh¹,

Kumar²,

Sultania³

et al. 2019

EAI Endorsed Transactions on Cloud Systems

View full text Add to dashboard Cite

show abstract

“…Hence, by integrating image processing techniques [6][7][8][9][10] and modifying the conventional method, [11,12] this study designs an automatic system for defects of printed art tiles, which is suitable for in-line inspection of printed tile defects. After analyzing the art tile texture features through gray level co-occurrence matrix, the analysis results were input into the backward propagation neural network in order to train the defect classifier.…”

Section: Introductionmentioning

confidence: 99%

Automatic Inspection System for Defects of Printed Art Tile Based on Texture Feature Analysis

Yang

Lin

et al. 2013

Instrumentation Science & Technology

View full text Add to dashboard Cite

& An automatic inspection system of printed art tile defects is reported in this artucke. After calculating eight defect features of art tile using the gray level co-occurrence matrix and the average R, G, B values of a defective area, the results were input into a backward propagation neural network for training the defect classifier. During inspection, the proposed system compared the inspected image with a standard image and removed noise by an erosion operation in order to preliminarily determine whether the art tile had defects. For the defective art tile images, the proposed classifier successfully identified four types of common printing defects. The proposed algorithm had an average recognition rate of 90%, suggesting that the recognition accuracy is good, and only requires 1-2 s to inspect an art tile with the size of 15 Â 15 cm 2 . The inspection speed is faster than the conventional manual inspection, and recognition results are more stable. The proposed system can reduce the risk of error caused by the long duration of manual inspection, and thus, can reduce manufacturing costs.

show abstract

“…The TFT-LCD polarizer plate attaching process uses the attaching roller of a PU (Polyurethane) surface material to evenly attach the two polarizer plates onto the glass substrate [1]. Lin et al listed many production defects that may occur which a ect the performance of the LCD product [2]. In the LCD polarizer plate attaching process, the prevalent problem of bubbles between the glass substrate and the polarizer plate, due to uneven attachment, often causes quality deviations in the LCD of color image light leakage.…”

Section: Introductionmentioning

confidence: 99%

Improvement of TFT-LCD's polarizer plate bubble problem using a preheating-process

Lin

Huang

et al. 2016

Scientia Iranica

View full text Add to dashboard Cite

In this study, improvement of the polarizer plate bubble in the thin lm transistor liquid crystal display process, especially where the polarizer plate is attached to the black matrix area at the end of the glass substrate, is presented. This study proposed temperature control functions during the attaching process using the preheating operations of pressure sensitive adhesives and a pre-existing warp polarizer plate. The polarizer plate is softened evenly, and, then, the bubbles can be discharged smoothly in the attaching process. Three main control factors are used in the polarizer plate attachment process, using an experimental design method, to determine the best combination of parameters. By applying optimal parameters in the experiments, bubble width can be decreased by 60.49%, compared with that of the previous process.

show abstract

High‐speed TFT LCD defect‐detection system with genetic algorithm

Cited by 17 publications

References 9 publications

Deep Learning based MURA Defect Detection

Deep Learning based MURA Defect Detection

Automatic Inspection System for Defects of Printed Art Tile Based on Texture Feature Analysis

Improvement of TFT-LCD's polarizer plate bubble problem using a preheating-process

Contact Info

Product

Resources

About