Research map / Image generation / Image restoration
Learning Disentangled Feature Representation for Hybrid-distorted Image Restoration
Xin Li, Xin Jin, Jianxin Lin, Sen Liu, Yaojun Wu, Tao Yu, Wei Zhou, Zhibo Chen
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 |
|---|---|---|---|
| add the feature disentanglement module (FDM) instead of plain ResBlocks (ablation) | FDR-Net · DID-HY | PSNR: 23.91 -> 24.97 | Improved |
| add the auxiliary ResBlock branch (ablation) | FDR-Net · DID-HY | PSNR: 24.27 -> 24.97 | Improved |
| increase number of FDlayers per FDM from 1 to 3 (ablation) | FDR-Net · DID-HY | PSNR: 24.63 -> 24.97 | Improved |
| increase FDlayers from 3 to 5 (ablation) | FDR-Net · DID-HY | PSNR: 24.97 -> 24.99 | No clear effect |
| increase channels from 16 to 32 (ablation) | FDR-Net · DID-HY | PSNR: 24.47 -> 24.97 | Improved |
| increase channels from 32 to 64 (ablation) | FDR-Net · DID-HY | PSNR: 24.97 -> 25.00 | No clear effect |
| increase phases from 2 to 6 (ablation) | FDR-Net · DID-HY | PSNR: 24.20 -> 24.97 | Improved |
| increase phases from 6 to 7 (ablation) | FDR-Net · DID-HY | PSNR: 24.97 -> 24.98 | No clear effect |
| apply FDM disentanglement to features | FDR-Net features · channel correlation | average channel correlation: 0.272 -> 0.164 | Improved |
| use FDR-Net instead of OP-Attention | hybrid restoration · DID-HY | PSNR: 23.1768 -> 24.9700 | Improved |
| apply FDR-Net to single distortion deraining instead of SPANET | FDR-Net · DID-MDN deraining | PSNR: 30.05 -> 32.96 | Improved |
Closest papers by meaning
- Learning Distortion Invariant Representation for Image Restoration From a Causality Perspective (CVPR 2023 · similarity 0.87)
- Boosting Filter Optimization and Prompt-Guided Decoding for Mixed Degradation Image Restoration (IJCV 2026 · similarity 0.85)
- Decouple to Reconstruct: High Quality UHD Restoration via Active Feature Disentanglement and Reversible Fusion (ICCV 2025 · similarity 0.85)
- Degradation-Aware Feature Perturbation for All-in-One Image Restoration (CVPR 2025 · similarity 0.84)
- CoDe: An Explicit Content Decoupling Framework for Image Restoration (CVPR 2024 · similarity 0.84)
- Correlation Matching Transformation Transformers for UHD Image Restoration (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.
