Research map / Image generation / Image restoration
Representing Noisy Image Without Denoising
Shuren Qi, Yushu Zhang, Chao Wang, Tao Xiang, Xiaochun Cao, Yong Xiang
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 |
|---|---|---|---|
| move moment computation from image domain to Radon domain (FM to FMR) | fractional-order moments · COREL, K=20, noise variance 0.2 | correct classification (%): 80.14 -> 95.72 | Improved |
| move moment computation from image domain to Radon domain (average over parameter settings) (ablation) | fractional-order moments · COREL, noise variance 0.2 | correct classification (%): 72.34 -> 85.02 | Improved |
| move moment computation to Radon domain, best results | fractional-order moments · COREL, noise variance 0.3 | correct classification (%): 45.31 -> 58.5 | Improved |
| move FJFM(3,3,2) to Radon domain | fractional-order moments · COREL, K=10, noise variance 0.3 | correct classification (%): 36.75 -> 33 | Hurt |
| use Radon projection for reconstruction instead of image domain (ablation) | harmonic FM · reconstruction of noisy Fourier image | MSRE: 0.060 -> 0.046 | Improved |
| add noise augmentation to rotation augmentation | ResNet-50 · COREL, noise variance 0.1 | correct classification (%): 1.28 -> 66.33 | Improved |
| add noise augmentation to rotation augmentation | GoogLeNet · COREL, noise variance 0.1 | correct classification (%): 2.22 -> 52.47 | Improved |
| add noise augmentation to rotation augmentation | PCANet · COREL, noise variance 0.1 | correct classification (%): 23.22 -> 13.72 | Hurt |
| add denoising preprocessing to rotation augmentation | ResNet-50 · COREL, noise variance 0.1 | correct classification (%): 1.28 -> 23.78 | Improved |
| add denoising preprocessing to rotation augmentation | ResNet-50 · COREL, noise variance 0.3 | correct classification (%): 1.25 -> 4 | Improved |
| add rotation augmentation to naive training | PCANet · COREL, clean images | correct classification (%): 19.64 -> 100 | Improved |
| use FMR instead of rotation-invariant SID | template matching · fluorescence microscopy, rotation and noise | accuracy (%): 20 -> 93 | Improved |
Closest papers by meaning
- Learning From Noisy Data With Robust Representation Learning (ICCV 2021 · similarity 0.87)
- Ditch the Denoiser: Emergence of Noise Robustness in Self-Supervised Learning from Data Curriculum (NeurIPS 2025 · similarity 0.86)
- Asymmetric Loss Functions for Noise-Tolerant Learning: Theory and Applications (TPAMI 2023 · similarity 0.85)
- On the Posterior Distribution in Denoising: Application to Uncertainty Quantification (ICLR 2024 · similarity 0.85)
- UNSURE: self-supervised learning with Unknown Noise level and Stein's Unbiased Risk Estimate (ICLR 2025 · similarity 0.83)
- Robust Visual Recognition with Class-Imbalanced Open-World Noisy Data (AAAI 2024 · similarity 0.83)
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.
