Research map / Image generation / Image restoration
Reblurring-Guided Single Image Defocus Deblurring: A Learning Framework with Misaligned Training Pairs
Dongwei Ren, Xinya Shu, Yu Li, Xiaohe Wu, Jin Hui Li, Wangmeng Zuo
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 |
|---|---|---|---|
| train with reblurring-guided framework (Eq. 1) | UNet · SDD | PSNR: 24.62 -> 25.82 | Improved |
| add defocus blur estimator and fusion block (Eq. 2) | UNet · SDD | PSNR: 25.82 -> 26.20 | Improved |
| train Loformer with reblurring framework (Eq. 1) | Loformer · SDD | PSNR: 26.51 -> 26.95 | Improved |
| add blur estimator and fusion block to Loformer (Eq. 2) | Loformer · SDD | PSNR: 26.51 -> 27.20 | Improved |
| add blur estimator and fusion block to Restormer | Restormer · SDD | PSNR: 26.89 -> 27.08 | Improved |
| add bi-directional optical flow deformation instead of L1 loss (ablation) | Ours (Eq. 1) · SDD | PSNR: 24.62 -> 25.82 | Improved |
| add cycle deformation (ablation) | Ours (Eq. 1) · SDD | PSNR: 24.80 -> 25.82 | Improved |
| add calibration mask (ablation) | Ours (Eq. 1) · SDD | PSNR: 25.78 -> 25.82 | Improved |
| add reblurring module (ablation) | Ours (Eq. 1) · SDD | PSNR: 25.58 -> 25.82 | Improved |
| add isotropic kernel prediction R_kpn (ablation) | Ours (Eq. 1) · SDD | PSNR: 25.68 -> 25.82 | Improved |
| add weight prediction network R_wpn (ablation) | Ours (Eq. 1) · SDD | PSNR: 25.74 -> 25.82 | Improved |
| use deformable cross-attention fusion instead of DConv (ablation) | UNet · SDD | PSNR: 26.01 -> 26.20 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning Single Image Defocus Deblurring with Misaligned Training Pairs (AAAI 2023 · similarity 0.96)
- Residual Diffusion Deblurring Model for Single Image Defocus Deblurring (AAAI 2025 · similarity 0.93)
- Learnable Blur Kernel for Single-Image Defocus Deblurring in the Wild (AAAI 2023 · similarity 0.93)
- Prior and Prediction Inverse Kernel Transformer for Single Image Defocus Deblurring (AAAI 2024 · similarity 0.91)
- Quad-Pixel Image Defocus Deblurring: A New Benchmark and Model (CVPR 2025 · similarity 0.91)
- LDP: Language-driven Dual-Pixel Image Defocus Deblurring Network (CVPR 2024 · 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.
