Research map / Image generation / Image restoration
Improving Lens Flare Removal with General-Purpose Pipeline and Multiple Light Sources Recovery
Yuyan Zhou, Dong Liang, Songcan Chen, Sheng-Jun Huang, Shuo Yang, Chongyi Li
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 |
|---|---|---|---|
| synthesize flare data with ISP-aware convex-combination pipeline instead of direct-add, Wu dataset | U-Net flare removal · Flare7K test set | PSNR: 23.6 -> 25.9 | Improved |
| synthesize flare data with convex-combination pipeline instead of direct-add, Flare7K training set | U-Net flare removal · Flare7K test set | PSNR: 25.4 -> 25.3 PSNR, 0.876 -> 0.884 SSIM | No clear effect |
| use Uformer with the proposed synthesis pipeline on Wu dataset | Uformer flare removal · Flare7K test set | SSIM: PSNR 23.7 -> 26.3, SSIM 0.863 -> 0.884 | Mixed |
| increase light-source weight exponent alpha from 1 to 15 (ablation) | light source recovery · Light source recovery | PSNR: 15.94 -> 17.88 | Improved |
| increase alpha beyond 15 (ablation) | light source recovery · Light source recovery | PSNR: 17.88 -> 17.88 | No clear effect |
| use soft power-weight light source recovery instead of hard illuminance threshold | flare removal post-processing · real images, nighttime and multiple lights | qualitative recovery | Improved |
| remove flare before running YOLOv5 detector | object detection · real flare images | qualitative detection | Improved |
| train with proposed synthesis pipeline on Wu data vs Wu direct-add (user study) | flare removal · user study, Wu test set | user preference: 55%:45% | Improved |
Closest papers by meaning
- Image Lens Flare Removal Using Adversarial Curve Learning (TPAMI 2025 · similarity 0.96)
- Light Source Guided Single-Image Flare Removal From Unpaired Data (ICCV 2021 · similarity 0.93)
- Disentangle Nighttime Lens Flares: Self-supervised Generation-based Lens Flare Removal (AAAI 2025 · similarity 0.92)
- FlareX: A Physics-Informed Dataset for Lens Flare Removal via 2D Synthesis and 3D Rendering (NeurIPS 2025 · similarity 0.91)
- Language-Guided One-Step Diffusion Model for Nighttime Flare Removal (CVPR 2026 · similarity 0.88)
- FLARE: Feed-forward Geometry, Appearance and Camera Estimation from Uncalibrated Sparse Views (CVPR 2025 · similarity 0.84)
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.
