Research map / Image generation / Image restoration
PBFG: A New Physically-Based Dataset and Removal of Lens Flares and Glares
Jie Zhu, Sungkil Lee
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 SRU to Uformer (ablation) | Uformer · Real-world nighttime (trained on PBFG) | PSNR: 29.124 -> 29.508 | Improved |
| add FEU to Uformer (ablation) | Uformer · Real-world nighttime (trained on PBFG) | PSNR: 29.124 -> 29.932 | Improved |
| combine SRU and FEU (full SFEM) (ablation) | Uformer · Real-world nighttime (trained on PBFG) | PSNR: 29.932 -> 30.366 | Improved |
| add SFEM to Uformer (ablation) | Uformer · Real-world nighttime (trained on Flare7K) | PSNR: 27.711 -> 29.330 | Improved |
| add histogram-matching module (ablation) | FGRNet · Real-world nighttime (trained on PBFG) | PSNR: 28.659 -> 30.366 | Improved |
| add perceptual loss Lvgg (ablation) | FGRNet · Real-world nighttime (trained on PBFG) | PSNR (dB): 29.330 -> 29.666 | Improved |
| add reconstruction loss Lrec (ablation) | FGRNet · Real-world nighttime (trained on PBFG) | PSNR (dB): 29.666 -> 30.053 | Improved |
| add negative SSIM loss (ablation) | FGRNet · Real-world nighttime (trained on PBFG) | PSNR (dB): 30.053 -> 30.366 | Improved |
| train on PBFG instead of Flare7K++ | Flare7K++ network · Real-world nighttime | PSNR: 27.316 -> 27.920 | Improved |
| train on PBFG instead of Flare7K | FGRNet · Real-world nighttime | PSNR: 28.228 -> 28.659 | Improved |
| train on PBFG instead of Flare7K++ | Wu et al. · Real-world nighttime | PSNR: 24.124 -> 25.601 | Improved |
| FEU alone instead of FEU plus SRU (ablation) | Uformer + FEU · Real-world nighttime | visual | Mixed |
Closest papers by meaning
- Flare7K++: Mixing Synthetic and Real Datasets for Nighttime Flare Removal and Beyond (TPAMI 2024 · similarity 0.91)
- Flare7K: A Phenomenological Nighttime Flare Removal Dataset (NeurIPS 2022 · similarity 0.91)
- Disentangle Nighttime Lens Flares: Self-supervised Generation-based Lens Flare Removal (AAAI 2025 · similarity 0.91)
- FlareX: A Physics-Informed Dataset for Lens Flare Removal via 2D Synthesis and 3D Rendering (NeurIPS 2025 · similarity 0.90)
- Language-Guided One-Step Diffusion Model for Nighttime Flare Removal (CVPR 2026 · similarity 0.87)
- RIFLE: Removal of Image Flicker-Banding via Latent Diffusion Enhancement (NeurIPS 2026 · 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.
