Research map / Image generation / Super-resolution
Correction Filter for Single Image Super-Resolution: Robustifying Off-the-Shelf Deep Super-Resolvers
Check 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 |
|---|---|---|---|
| Apply the known-kernel correction filter to bicubic-trained DBPN | DBPN · Set14, non-blind, Gaussian kernels and box kernels at scales 2 and 4 | PSNR (dB): 30.078 -> 34.023; 26.366 -> 33.288; 28.444 -> 29.364; 25.067 -> 28.680; 23.890 -> 28.267; 24.636 -> 25.157 | Helped |
| Apply the known-kernel correction filter to bicubic-trained DBPN | DBPN · BSD100, non-blind, Gaussian kernels at scales 2 and 4 | PSNR (dB): 29.512 -> 32.300; 26.371 -> 31.875; 25.268 -> 27.690; 24.357 -> 27.474 | Helped |
| Estimate and apply a correction filter without knowing the kernel | DBPN · Set14, blind, Gaussian std 1.5/√2 | PSNR (dB): 30.078 -> 28.46 | Hurt |
| Estimate and apply a correction filter without knowing the kernel | DBPN · Set14, blind, Gaussian std 2.5/√2 and Box width 4 | PSNR (dB): 26.366 -> 28.037; 28.444 -> 29.778 | Helped |
| Estimate and apply a correction filter without knowing the kernel | DBPN · BSD100, blind, scale 2 Gaussian std 2.5/√2 | PSNR (dB): 26.371 -> 27.761; KernelGAN: 28.244 | Mixed |
| Mimic bicubic-kernel LR observations with a correction filter before a pretrained super-resolver | DBPN, RCAN, and ProSR · Non-blind Set14 and BSD100 | PSNR (dB) | Helped |
Closest papers by meaning
- Deep Constrained Least Squares for Blind Image Super-Resolution (CVPR 2022 · similarity 0.90)
- Closed-Loop Matters: Dual Regression Networks for Single Image Super-Resolution (CVPR 2020 · similarity 0.90)
- Blind Super-Resolution With Iterative Kernel Correction (CVPR 2019 · similarity 0.90)
- Learning a Single Convolutional Super-Resolution Network for Multiple Degradations (CVPR 2018 · similarity 0.90)
- Noise-Free Optimization in Early Training Steps for Image Super-resolution (AAAI 2024 · similarity 0.89)
- A Dynamic Kernel Prior Model for Unsupervised Blind Image Super-Resolution (CVPR 2024 · similarity 0.88)
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.