Automatic TFT-LCD mura detection based on image reconstruction and processing

Automatic inspection of Mura defects is a challenging task in thin-film transistor liquid crystal display (TFT-LCD) defect detection, which is critical for LCD manufacturers to guarantee high standard quality control. In this paper, we propose a set of automatic procedures to detect mura defects by using image processing and computer vision techniques. Singular Value Decomposition (SVD) and Discrete Cosine Transformation(DCT) techniques are employed to conduct image reconstruction, based on which we are able to obtain the differential image of LCD Cells. In order to detect different types of mura defects accurately, we then design a method that employs different detection modules adaptively, which can overcome the disadvantage of simply using a single threshold value. Finally, we provide the experimental results to validate the effectiveness of the proposed method in mura detection.