State of the Art on Demoiréing Models

초록

Moiré patterns typically occur when capturing objects with complex textures or when photographing screens using a camera. These patterns arise because of frequency aliasing, which is caused by structural mismatch between the sensor of the imaging device and the pixel arrangement of the screen. These patterns obscure the original data and cause color distortions, thus significantly degrading the performance of downstream tasks when such images are applied. Various demoiréing methods have been proposed to address these challenges. In this study, we conduct a comprehensive survey of demoiréing models. Specifically, we examine models designed for both image and video domains. Subsequently, we highlight approaches that utilize the spatial and frequency domains, as well as those leveraging standard RGB (sRGB) and RAW domains. Furthermore, we provide an overview of the metrics and datasets employed in these models to evaluate their performances.

키워드