Research map / Image generation / Image restoration
Relationship Quantification of Image Degradations
Boyun Li, Yuanbiao Gou, Wenxin Wang, Peng Hu, Wangmeng Zuo, Xi Peng
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 10% denoising data to dehazing training (ablation) | MSBDN · RESIDE+ outdoor dehazing | PSNR: 33.84 -> 33.96 | Improved |
| raise noise proportion to 90% (ablation) | MSBDN · RESIDE+ outdoor dehazing | PSNR: 33.84 -> 30.96 | Hurt |
| raise noise proportion to 30% (ablation) | MSBDN · RESIDE+ outdoor dehazing | PSNR: 33.84 -> 33.00 | Hurt |
| raise noise proportion from 10% to 20% (ablation) | MSBDN · RESIDE+ outdoor dehazing | PSNR: 33.96 -> 33.93 | No clear effect |
| add 10% rain data to dehazing training | MSBDN · SOTS+ indoor | PSNR: 32.72 -> 35.14 | Improved |
| add 10% rain data to dehazing training | Dehamer · SOTS+ outdoor | PSNR: 35.18 -> 36.11 | Improved |
| add 10% snow data to dehazing training | AECR-Net · SOTS+ indoor | PSNR: 37.17 -> 37.29 | No clear effect |
| add auxiliary degradations (noise, rain, snow) at 10% averaged over seven baselines | seven dehazing baselines · SOTS+ | PSNR: +0.42dB, +0.77dB, +0.63dB | Improved |
| add 10% haze auxiliary to rain restoration | MSBDN · SOTS+ rain anchor | PSNR: 38.12 -> 36.66 | Hurt |
| add 10% rain auxiliary to noise restoration | MSBDN · SOTS+ noise anchor | PSNR: 34.85 -> 34.83 | No clear effect |
| use DRI instead of Taskonomy affinity to predict benefit of 10% noise | haze anchor MSBDN · RESIDE+ haze | sign of predictor vs PSNR gain: -1.00000 -> 0.00860 | Improved |
| compute DRI every 5 steps instead of every step (ablation) | DRI · Pearson with final performance | Pearson coefficient: 0.9698 -> 0.9721 | Improved |
Closest papers by meaning
- Unsupervised Degradation Representation Aware Transform for Real-World Blind Image Super-Resolution (AAAI 2025 · similarity 0.85)
- Visual-Instructed Degradation Diffusion for All-in-One Image Restoration (CVPR 2025 · similarity 0.84)
- DDR: Exploiting Deep Degradation Response as Flexible Image Descriptor (NeurIPS 2024 · similarity 0.84)
- Degradation-Aware Feature Perturbation for All-in-One Image Restoration (CVPR 2025 · similarity 0.84)
- Learning Where and What to Restore for Composite Image Restoration (NeurIPS 2026 · similarity 0.84)
- Reti-Diff: Illumination Degradation Image Restoration with Retinex-based Latent Diffusion Model (ICLR 2025 · similarity 0.84)
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.
