Research map / Image generation / Image restoration
Deep Halftoning With Reversible Binary Pattern
Menghan Xia, Wenbo Hu, Xueting Liu, Tien-Tsin Wong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add noise incentive block to the dithering network (halftone PSNR) (ablation) | Reversible halftoning · VOC2012 test (color) | PSNR: 31.915 -> 33.734 | Improved |
| add noise incentive block to the dithering network (halftone SSIM) (ablation) | Reversible halftoning · VOC2012 test (color) | SSIM: 0.1514 -> 0.1702 | Improved |
| add noise incentive block (restoration PSNR) (ablation) | Reversible halftoning · VOC2012 test (color restoration) | PSNR: 27.743 -> 29.112 | Improved |
| add noise incentive block (restoration SSIM) (ablation) | Reversible halftoning · VOC2012 test (color restoration) | SSIM: 0.8667 -> 0.8826 | Improved |
| add NIB to invertible grayscale model IG | IG · DIV2K | PSNR: 38.411 -> 39.314 | Improved |
| add NIB to invertible grayscale model IG | IG · DIV2K | SSIM: 0.9765 -> 0.9811 | Improved |
| add NIB to SPADE | SPADE · Cityscapes | FID: 71.8 -> 54.6 | Improved |
| add NIB to SPADE | SPADE · Cityscapes | mIOU: 62.3% -> 61.9% | No clear effect |
| add NIB to SPADE | SPADE · Cityscapes | pixel accuracy: 81.9% -> 81.8% | No clear effect |
| add blue-noise loss to remove stripe patterns (ablation) | Reversible halftoning · qualitative (Figure 4) | visual pattern artifacts | Improved |
| add binarization loss (ablation) | Reversible halftoning · training stability | training convergence | Improved |
| add guidance loss in warm-up stage (ablation) | Reversible halftoning · warm-up training | convergence speed: 28 epochs vs 130 epochs | Improved |
Closest papers by meaning
- Conditional Diffusion Process for Inverse Halftoning (NeurIPS 2022 · similarity 0.88)
- Ink Dot-Oriented Differentiable Optimization for Neural Image Halftoning (CVPR 2024 · similarity 0.88)
- Deep Restoration of Vintage Photographs From Scanned Halftone Prints (ICCV 2019 · similarity 0.85)
- Colorful Image Colorization (ECCV 2016 · similarity 0.83)
- Quasi-Unsupervised Color Constancy (CVPR 2019 · similarity 0.83)
- Distort-and-Recover: Color Enhancement Using Deep Reinforcement Learning (CVPR 2018 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
