Research map / Image generation / Image restoration
UHD-MFF: Shattering Barriers in Multi-Focus Ultra-High-Definition Image Fusion via Learnable Lookup Tables
Yibing Zhang, Xunpeng Yi, Qinglong Yan, Yeda Wang, Han Xu, Jiayi Ma
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 dilated gradient cue to the C-LUT (ablation) | UMF-LUT · UHD-MFF-Syn | VIF: 1.2564 -> 1.4335 | Improved |
| add gradient difference cue to the C-LUT (ablation) | UMF-LUT · UHD-MFF-Syn | VIF: 1.4028 -> 1.4335 | Improved |
| add maximum gradient cue to the C-LUT (ablation) | UMF-LUT · UHD-MFF-Syn | VIF: 1.3948 -> 1.4335 | Improved |
| add learnable regional semantic cue to the C-LUT (ablation) | UMF-LUT · UHD-MFF-Syn | spatial frequency: 5.8832 -> 6.7443 | Improved |
| add the Detail-Edge LUT branch (ablation) | UMF-LUT · UHD-MFF-Syn | average gradient: 1.8479 -> 1.9416 | Improved |
| train on 1500 pairs instead of 15 (ablation) | UMF-LUT · UHD-MFF-Syn | Q^AB/F: 0.5367 -> 0.7498 | Improved |
| train on 1500 pairs instead of 150 (ablation) | UMF-LUT · UHD-MFF-Syn | Q^AB/F: 0.5400 -> 0.7498 | Improved |
| train on native 4K data instead of low-resolution data (ablation) | UMF-LUT · UHD-MFF-Syn | Q^AB/F: 0.6129 -> 0.7498 | Improved |
| use unsupervised loss instead of supervised loss (ablation) | UMF-LUT · UHD-MFF-Syn | Q^AB/F: 0.7491 -> 0.7498 | No clear effect |
| use UMF-LUT instead of CCSR-Net | UHD-MFF-Syn fusion · UHD-MFF-Syn | SD: 49.8757 -> 49.9062 | Improved |
| use UMF-LUT instead of TC-MoA | UHD-MFF-Syn fusion · UHD-MFF 4K efficiency | test time (ms): 5714.91 -> 11.03 | Improved |
Closest papers by meaning
- Ultra-High-Definition Dynamic Multi-Exposure Image Fusion via Infinite Pixel Learning (AAAI 2025 · similarity 0.86)
- Hybrid-Supervised Dual-Search: Leveraging Automatic Learning for Loss-Free Multi-Exposure Image Fusion (AAAI 2024 · similarity 0.85)
- Multi-Focus Image Fusion via Explicit Defocus Blur Modelling (AAAI 2025 · similarity 0.84)
- UltraFusion: Ultra High Dynamic Imaging using Exposure Fusion (CVPR 2025 · similarity 0.84)
- Ultra-High Resolution Segmentation via Boundary-Enhanced Patch-Merging Transformer (AAAI 2025 · similarity 0.84)
- UltraHR-100K: Enhancing UHR Image Synthesis with A Large-Scale High-Quality Dataset (NeurIPS 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.
