Research map / Image generation / Image restoration
Physics-based Feature Dehazing Networks
Jiangxin Dong, Jinshan Pan
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 FDU in place of plain features (ablation) | PFDN (9 ResBlocks baseline) · NYU & Make3D synthetic test | PSNR (dB): 26.33 -> 27.26 | Improved |
| add FDU instead of stacking 15 ResBlocks (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 26.75 -> 27.26 | Improved |
| increase ResBlocks from 9 to 15 without FDU (ablation) | PFDN w/o FDU · NYU & Make3D synthetic test | PSNR (dB): 26.33 -> 26.75 | Improved |
| add residual learning in PFDB (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 23.55 -> 27.26 | Improved |
| add residual learning in PFDB (ablation) | PFDN · NYU & Make3D synthetic test | SSIM: 0.7935 -> 0.9047 | Improved |
| add GAP in the FDU (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 26.56 -> 27.26 | Improved |
| replace channel attention with FDU (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 26.64 -> 27.26 | Improved |
| replace channel attention with FDU (ablation) | PFDN · NYU & Make3D synthetic test | SSIM: 0.8846 -> 0.9047 | Improved |
| increase number of PFDBs from 9 to 15 (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 27.26 -> 27.58 | Improved |
| increase number of PFDBs from 15 to 17 (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 27.58 -> 27.33 | Hurt |
| reduce number of PFDBs from 9 to 7 (ablation) | PFDN · NYU & Make3D synthetic test | PSNR (dB): 27.26 -> 27.17 | No clear effect |
| PFDN versus GDN | GDN · RESIDE SOTS indoor | SSIM: 0.9836 -> 0.9760 | Hurt |
Closest papers by meaning
- Advancing Real-World Image Dehazing: Perspective, Modules, and Training (TPAMI 2024 · similarity 0.88)
- Towards Perceptual Image Dehazing by Physics-Based Disentanglement and Adversarial Training (AAAI 2018 · similarity 0.87)
- FFA-Net: Feature Fusion Attention Network for Single Image Dehazing (AAAI 2020 · similarity 0.87)
- Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network (NeurIPS 2024 · similarity 0.86)
- Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network (AAAI 2025 · similarity 0.85)
- PEIE: Physics Embedded Illumination Estimation for Adaptive Dehazing (AAAI 2025 · similarity 0.85)
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.
