Defect detection of four-color display screen based on color equalization and local dynamic threshold segmentation

Konferenz: AIIPCC 2022 - The Third International Conference on Artificial Intelligence, Information Processing and Cloud Computing
21.06.2022 - 22.06.2022 in Online

Tagungsband: AIIPCC 2022

Seiten: 5Sprache: EnglischTyp: PDF

Autoren:
Shuai, Lingyu; Chen, Huaixin (School of Resource and Environment, University of Electronic Science and Technology of China, Sichuan, China)
Wang, Zhixi (Novel Product R & D Department, Truly Opto-Electronics Co., Ltd., Shanwei, China)

Inhalt:
It is easy to have a range of flaws in the TFT-LCD manufacturing process. As a typical detection image, four-color detection image can identify a range of faults as a standard detection screen. This work proposes a display defect detection approach based on color equalization and local threshold segmentation for four-color detection images. The input image is first preprocessed. The input image is then sent into an SVM for color deviation detection. When the input image has defects other than color deviation, convert it to HSI color space and calculate the image's histogram, smooth the histogram curve, and calculate the gray scale's maximum trough point. The color balanced image is then obtained by applying the adaptive strain interval γ transform to the image histogram. The binary image is then created using the Sauvola local threshold segmentation approach. Finally, post-processing is used to obtain the defect detection binary image, which completes the defect detection. The suggested approach can detect display screen defects in multi-color backgrounds and entirely segment the defective region, which has a wide range of applications.