Multiscale Coarse-to-Fine Guided Screenshot Demoiréing

Duong Hai Nguyen1, Se-Ho Lee2, and Chul Lee1
1Department of Multimedia Engineering, Dongguk University, South Korea
2Department of Information and Engineering, Jeonbuk National University, South Korea
IEEE Signal Processing Letters 2023
Paper Code

Qualitative comparisons of demoireing results on the UHDM dataset.
Given a moiré image, from left to right, top to bottom, the magnified views show the results of DMCNN, MDDM, WDNet, MopNet, MBCNN, FHDe²Net, ESDNet, ours, and ground truth.

Abstract

In this letter, we propose a multiscale coarse-to-fine guided screenshot demoiréing algorithm. We first extract the multiscale feature of the input image. Then, we develop the multiscale guided restoration block (MGRB), which removes moiré patterns with the guidance of multiscale information by exploiting the correlation between moiré frequencies. To this end, we design two blocks for feature modulation and moiré pattern removal. In addition, to further improve the performance, we develop an adaptive reconstruction loss to direct the network to focus on regions that are difficult to restore. Experimental results on multiple datasets demonstrate that the proposed algorithm provides comparable or even better demoiréing performance than state-of-the-art algorithms.

Network Architecture & Quantitative Comparison

BibTeX


      
@article{2023_nguyen_gad,
  author  = {Duong Hai, Nguyen and Se-Ho, Lee and Chul, Lee},
  title   = {Multiscale Coarse-to-Fine Guided Screenshot Demoiréing},
  journal = {IEEE Signal Processing Letters},
  volume  = {30},
  number  = {},
  pages   = {898-902},
  month   = jul,
  year    = {2023},
  doi     = {10.1109/LSP.2023.3296039}
}