Research map / Image generation / Image restoration
Blind Image Deblurring with Unknown Kernel Size and Substantial Noise
Zhong Zhuang, Taihui Li, Hengkang Wang, Ju Sun
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 l1/l2 instead of l1 regularizer on image gradient (ablation) | double-DIP blind deblurring · Levin, low Gaussian noise | PSNR (dB): 31.12 -> 32.64 | Improved |
| use l1/l2 instead of l1 regularizer on image gradient (ablation) | double-DIP blind deblurring · Levin, high Gaussian noise | PSNR (dB): 24.34 -> 27.74 | Improved |
| use l1/l2 so one lambda works across noise levels (ablation) | double-DIP blind deblurring · Levin, Gaussian noise | optimal lambda_x: 0.0001 -> 0.0002 | Improved |
| replace MLP-based DIP for the kernel with SIREN (ablation) | double-DIP blind deblurring · synthetic noisy blur | kernel and image quality | Improved |
| add windowed-moving-variance early stopping | double-DIP blind deblurring · high Gaussian noise | PSNR curve | Improved |
| increase regularization strength instead of early stopping (ablation) | double-DIP blind deblurring · Gaussian noise sigma=0.05 | PSNR curve | No clear effect |
| overspecify kernel size up to half the image size | proposed method vs other single-instance methods · LEVIN11 and LAI16 | PSNR, SSIM, VIF, LPIPS | Improved |
| use Huber robust loss instead of MSE | double-DIP blind deblurring · LAI16 impulse, shot noise | PSNR, SSIM, VIF, LPIPS | Improved |
Closest papers by meaning
- Cross-Scale Self-Supervised Blind Image Deblurring via Implicit Neural Representation (NeurIPS 2024 · similarity 0.90)
- Non-Blind Deblurring: Handling Kernel Uncertainty With CNNs (CVPR 2018 · similarity 0.90)
- Learning a Discriminative Prior for Blind Image Deblurring (CVPR 2018 · similarity 0.89)
- Blind image deblurring with noise-robust kernel estimation (ECCV 2024 · similarity 0.89)
- Uncertainty-Aware Unsupervised Image Deblurring With Deep Residual Prior (CVPR 2023 · similarity 0.89)
- Unsupervised Blind Image Deblurring Based on Self-Enhancement (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.
