Research map / Image generation / Super-resolution
Image Formation Model Guided Deep Image Super-Resolution
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 cascaded pixel-substitution image-formation constraint | Deep SR network · Set5 ×4 | PSNR: 32.48 -> 32.56 | Helped |
| Add cascaded pixel-substitution image-formation constraint | Deep SR network · Urban100 ×4 | PSNR: 26.65 -> 26.72 | Helped |
| Add cascaded pixel-substitution image-formation constraint | Deep SR network · Set14 ×4 | PSNR: 28.82 -> 28.80 | Hurt |
| Remove convolution and pixel substitution (ablation) | Cascaded SR network · Set5 ×2 | PSNR: 38.26 -> 38.20 | Hurt |
| Use only one feed-forward stage instead of the cascade (ablation) | Cascaded SR network · Set5 ×2 | PSNR: 38.26 -> 38.19 | Hurt |
| Replace pixel-substitution feedback with an image-formation loss (ablation) | One-stage SR network · Set5 ×2 | PSNR: 38.26 -> 38.22 | Hurt |
| Add image-formation loss to the one-stage network (ablation) | One-stage SR network · Set5 ×2 | PSNR: 38.19 -> 38.22 | Helped |
| Continue cascading from Stage 1 to Stage 2 (ablation) | Proposed SR network · Set5 ×2 | PSNR: 38.23 -> 38.27 | Helped |
| Continue cascading from Stage 2 to Stage 3 (ablation) | Proposed SR network · Set5 ×2 | PSNR: 38.27 -> 38.26 | Hurt |
| Use pixel-substitution feedback rather than regenerated LR feedback | SR network · Set5 ×2 regenerated LR | PSNR: 61.60 -> 72.06 | Helped |
| Apply the assumed SR image-formation model to JPEG-compressed inputs | Proposed SR method · Images with significant JPEG compression artifacts | artifact severity | Hurt |
Closest papers by meaning
- Accurate Image Super-Resolution Using Very Deep Convolutional Networks (CVPR 2016 · similarity 0.89)
- Correction Filter for Single Image Super-Resolution: Robustifying Off-the-Shelf Deep Super-Resolvers (CVPR 2020 · similarity 0.88)
- Learning a Single Convolutional Super-Resolution Network for Multiple Degradations (CVPR 2018 · similarity 0.88)
- Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network (CVPR 2016 · similarity 0.88)
- ResDiff: Combining CNN and Diffusion Model for Image Super-resolution (AAAI 2024 · similarity 0.88)
- Noise-Free Optimization in Early Training Steps for Image Super-resolution (AAAI 2024 · similarity 0.87)
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.