Research map / Image generation / Image restoration
NBNet: Noise Basis Learning for Image Denoising With Subspace Projection
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 |
|---|---|---|---|
| Use NBNet with SSA-based feature projection | Image denoising on SIDD · SIDD | PSNR: 39.72 -> 39.75 | Helped |
| Use NBNet with SSA-based feature projection | Image denoising on SIDD · SIDD | SSIM: 0.959 -> 0.973 | Helped |
| Use NBNet with SSA-based feature projection | Image denoising on DND · DND | PSNR: 39.88 -> 39.89 | Helped |
| Use NBNet with SSA-based feature projection | Image denoising on DND · DND | SSIM: 0.956 -> 0.955 | Hurt |
| Add SSA to UNet (ablation) | UNet on SIDD · SIDD | PSNR: 39.62 -> 39.68 | Helped |
| Add convolutional blocks to UNet (ablation) | UNet on SIDD · SIDD | PSNR: 39.62 -> 39.69 | Helped |
| Add SSA to UNet with convolutional blocks (ablation) | UNet on SIDD · SIDD | PSNR: 39.69 -> 39.75 | Helped |
| Concatenate first- and penultimate-layer features (ablation) | DnCNN on SIDD · SIDD | PSNR: 38.04 -> 38.21 | Helped |
| Add SSA using first- and penultimate-layer features (ablation) | DnCNN on SIDD · SIDD | PSNR: 38.04 -> 38.59 | Helped |
| Set subspace dimension K to 1 instead of 16 (ablation) | NBNet on SIDD · SIDD | PSNR: 39.75 -> 39.28 | Hurt |
| Set subspace dimension K to 8 instead of 16 (ablation) | NBNet on SIDD · SIDD | PSNR: 39.75 -> 39.74 | No clear effect |
| Set subspace dimension K to 32 (ablation) | NBNet on SIDD · SIDD | convergence | Hurt |
Closest papers by meaning
- Zero-Shot Noise2Noise: Efficient Image Denoising Without Any Data (CVPR 2023 · similarity 0.88)
- Noisier2Noise: Learning to Denoise From Unpaired Noisy Data (CVPR 2020 · similarity 0.87)
- Next-Scale Prediction: A Self-Supervised Approach for Real-World Image Denoising (CVPR 2026 · similarity 0.87)
- APR-RD: Complemental Two Steps for Self-Supervised Real Image Denoising (AAAI 2025 · similarity 0.87)
- Zero-shot Denoising via Neural Compression: Theoretical and algorithmic framework (NeurIPS 2025 · similarity 0.86)
- Zero-Shot Noise2Mean: Gap Minimization for Efficient Denoising from a Single Noisy Image (AAAI 2025 · similarity 0.86)
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.