Underexposed Image Enhancement for Mura Compensation in Display Panels

Abstract

Mura, a form of uneven brightness on display panels, manifests as visible artifacts that compromise display quality. Conventional mura compensation methods rely on long-exposure imaging to accurately capture subtle brightness inconsistencies, but this approach is time-intensive and unsuitable for high-throughput manufacturing. In this study, we propose a novel method that leverages underexposed image enhancement to estimate mura compensation data using short-exposure imaging. Specifically, we introduce a novel mura compensation loss function designed to minimize discrepancies in compensation data derived from long- and short-exposure images. To facilitate supervised training, we construct a dataset comprising paired long- and short-exposure images of display panels with real and synthesized mura patterns. Our experimental results demonstrate the effectiveness of the proposed approach, showing that the enhancement network trained with the proposed loss function achieves precise mura compensation while significantly reducing imaging time, offering a practical solution for efficient manufacturing processes.