Coarse-to-fine Disentangling Demoiring Framework for Recaptured Screen Images

Ce Wang,Bin He,Shengsen Wu,Renjie Wan,Boxin Shi,Ling-Yu Duan

DOI: https://doi.org/10.1109/TPAMI.2023.3243310

IF: 23.6

2023-01-01

IEEE Transactions on Pattern Analysis and Machine Intelligence

Abstract:Removing the undesired moire patterns from images capturing the contents displayed on screens is of increasing research interest, as the need for recording and sharing the instant information conveyed by the screens is growing. Previous demoireing methods provide limited investigations into the formation process of moire patterns to exploit moire-specific priors for guiding the learning of demoireing models. In this paper, we investigate the moire pattern formation process from the perspective of signal aliasing, and correspondingly propose a coarse-to-fine disentangling demoireing framework. In this framework, we first disentangle the moire pattern layer and the clean image with alleviated ill-posedness based on the derivation of ourmoire image formation model. Then we refine the demoireing results exploiting both the frequency domain features and edge attention, considering moire patterns' property on spectrum distribution and edge intensity revealed in our aliasing based analysis. Experiments on several datasets showthat the proposedmethod performs favorably against state-of-the-art methods. Besides, the proposed method is validated to adapt well to different data sources and scales, especially on the high-resolution moire images.