Research map / Image generation / Image restoration
Revisiting Image Fusion for Multi-Illuminant White-Balance Correction
David Serrano-Lozano, Aditya Arora, Luis Herranz, Konstantinos G. Derpanis, Michael S. Brown, Javier Vazquez-Corral
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 transformer-based non-linear fusion of five WB presets | DeepWB · Multi-illuminant dataset (Sony+Nikon) | mean DeltaE2000: 6.83 -> 4.55 | Improved |
| replace linear per-pixel blending with transformer fusion | MixedWB · Multi-illuminant dataset (Sony+Nikon) | mean DeltaE2000: 9.31 -> 4.55 | Improved |
| replace linear blending with transformer fusion | StyleWB · Multi-illuminant dataset (Sony split) | mean DeltaE2000: 9.46 -> 4.71 | Improved |
| use all five WB presets instead of daylight only (ablation) | Ours · Multi-illuminant dataset (Sony+Nikon) | mean DeltaE2000: 5.47 -> 4.55 | Improved |
| use all five presets instead of three (DST) (ablation) | Ours · Multi-illuminant dataset (Sony+Nikon) | mean DeltaE2000: 5.18 -> 4.55 | Improved |
| use all five presets instead of daylight only (ablation) | Ours · Multi-illuminant dataset (Sony+Nikon) | mean MAE: 4.27 -> 3.61 | Improved |
| use linear fusion of presets (MixedWB) instead of standard sRGB WB (DeepWB) | DeepWB · Multi-illuminant dataset (Sony+Nikon) | mean DeltaE2000: 6.83 -> 9.31 | Hurt |
| transformer fusion trained on Nikon, tested on Sony | MixedWB · Cross-camera (train Nikon, test Sony) | mean DeltaE2000: 9.17 -> 5.80 | Improved |
| transformer fusion on synthetic multi-illuminant test set | StyleWB · Cross-dataset (Synthetic test set) | mean DeltaE2000: 9.90 -> 7.53 | Improved |
| transformer fusion on single-illuminant Cube+ | MixedWB · Cube+ single illuminant | DeltaE: 5.03 -> 4.19 | Improved |
| transformer fusion with channel-wise transposed attention | DeepWB · inference time, 50 images 500x700 | runtime: 459 -> 179 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SWBNet — SWBNet: A Stable White Balance Network for sRGB Images (AAAI 2023) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Large Scale Multi-Illuminant (LSMI) Dataset for Developing White Balance Algorithm Under Mixed Illumination (ICCV 2021 · similarity 0.91)
- Attentive Illumination Decomposition Model for Multi-Illuminant White Balancing (CVPR 2024 · similarity 0.89)
- ABC-Former: Auxiliary Bimodal Cross-domain Transformer with Interactive Channel Attention for White Balance (CVPR 2025 · similarity 0.87)
- White-Balance First, Adjust Later: Cross-Camera Color Constancy via Vision-Language Evaluation (CVPR 2026 · similarity 0.86)
- HVDualformer: Histogram-Vision Dual Transformer for White Balance (AAAI 2025 · similarity 0.86)
- Enhancing Low-Light Images: A Synthetic Data Perspective on Practical and Generalizable Solutions (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.
