Research map / Image generation / Image restoration
Non-Uniform Blind Deblurring by Reblurring
Yuval Bahat, Netalee Efrat, Michal Irani
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 |
|---|---|---|---|
| use training-free reblurring blur-field and patch-recurrence deblurring instead of CNN-based method [19] | Ours · Non-uniform deblurring (moderate blur, up to 16 px) | PSNR (dB): +0.73dB | Improved |
| use training-free reblurring blur-field and patch-recurrence deblurring instead of CNN-based method [19] | Ours · Non-uniform deblurring (large blur, up to 38 px) | PSNR (dB): +1.2dB | Improved |
| use our method instead of [23] 3D camera motion model | Ours · Non-uniform deblurring (large blur, up to 38 px) | PSNR (dB): +1.54dB | Improved |
| use our method instead of CNN-based [19] for flow estimation | Ours · Non-uniform deblurring (large blur, up to 38 px) | flow error (EPE, pixels): 8.64 -> 5.74 | Improved |
| use our flow estimation instead of CNN-based [19] | Ours · Non-uniform deblurring (moderate blur, up to 16 px) | flow error (EPE, pixels): 2.27 -> 3.19 | Hurt |
| use our method instead of [19] on a large synthetic blur example | Ours · Figure 2 large synthetic blur example | PSNR (dB): 15.72 -> 18.51 | Improved |
| use locally linear blur model on highly non-linear camera-motion blurs | Ours · Non-linear blur dataset of [11] | mean PSNR (dB): 26.33dB | No clear effect |
Closest papers by meaning
- Non-Blind Deblurring: Handling Kernel Uncertainty With CNNs (CVPR 2018 · similarity 0.90)
- Blind Image Deblurring With Local Maximum Gradient Prior (CVPR 2019 · similarity 0.89)
- Unsupervised Blind Image Deblurring Based on Self-Enhancement (CVPR 2024 · similarity 0.89)
- Deblurring Natural Image Using Super-Gaussian Fields (ECCV 2018 · similarity 0.89)
- A Neural Approach to Blind Motion Deblurring (ECCV 2016 · similarity 0.89)
- End-to-end Interpretable Learning of Non-blind Image Deblurring (ECCV 2020 · similarity 0.89)
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.
