TFT-LCD Mura Defect Detection Based on ICA and Multi-channels Fusion

Mura defect is a kind of common defects in thin-film-transistor liquid-crystal display (TFT-LCD). This paper proposes an effective detection method for mura defects based on background reconstruction and multi-channels fusion detection. Independent component analysis (ICA) is used to get a group of basis images from a set of faultless images for background reconstruction which will reserve the background information as much as possible and is not affected by target region at the same time. A multi-thresholds segmentation model which can reduce noise and reserve mura regions simultaneously is proposed to segment the difference image which is calculated by subtracting the reconstructed back-ground from the TFT-LCD images to be detected. Considering the color information of mura defects, multi-channels detection method is constructed to detect mura defects both in color space and gray space. The experimental results show that the proposed method can detect different types of mura defects accurately.