Research map / Image generation / Super-resolution
CutMIB: Boosting Light Field Super-Resolution via Multi-View Image Blending
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train existing light-field SR networks with CutMIB instead of without it | ATO, InterNet, IINet, DPT, and DistgSSR · Light-field SR, ×2 and ×4; five benchmark datasets | PSNR: +0.007 to +0.395 | Helped |
| Train InterNet with CutMIB rather than CutBlur | InterNet · Light-field SR ×4, STFgantry | PSNR: 0.319 dB gain compared with training using CutBlur | Helped |
| Use all four directional view groups rather than only θ1 (ablation) | InterNet with CutMIB · HCInew ×4 | PSNR: 30.942 -> 31.002 | Helped |
| Increase the number of selected views from k = 1 to k = 20 (ablation) | InterNet with CutMIB · HCInew ×4 | PSNR: 30.942 -> 31.006 | Helped |
| Use k = 1 view rather than the default all-view blending (ablation) | InterNet with CutMIB · INRIA ×4 | PSNR: 30.743 -> 30.736 | Hurt |
| Train with CutMIB | IINet · INRIA ×2 and ×4 | PSNR: -0.026; -0.018 at ×2, -0.060 at ×4 | Mixed |
| Train with CutMIB | DistgSSR · Light-field SR ×2 and ×4; five benchmark datasets | PSNR: -0.025 to +0.020 at ×2; +0.003 to +0.046 at ×4 | Mixed |
| Train existing networks with CutMIB | ATO, InterNet, and IINet · Light-field denoising, σ = 10, 20, 50 | PSNR: +0.013 to +0.232 | Helped |
| Train SR networks with CutMIB | ATO, IINet, InterNet, and DPT · EPFL ×4 with isotropic Gaussian kernels k = 1.8, 2.5, 3.2 | PSNR: 0.000 to +0.075 | Helped |
| Use CutMIB during training | Existing light-field SR networks · Real-world light-field SR | visual reconstruction quality | Helped |
| Use CutMIB during training | Existing light-field SR networks · Angular consistency of reconstructed light fields | MSE | Helped |
Closest papers by meaning
- Rethinking Data Augmentation for Image Super-resolution: A Comprehensive Analysis and a New Strategy (CVPR 2020 · similarity 0.90)
- NeLF-Pro: Neural Light Field Probes for Multi-Scale Novel View Synthesis (CVPR 2024 · similarity 0.84)
- Exploiting Deblurring Networks for Radiance Fields (CVPR 2025 · similarity 0.84)
- GSDD: Generative Space Dataset Distillation for Image Super-resolution (AAAI 2024 · similarity 0.84)
- Continuous Optical Zooming: A Benchmark for Arbitrary-Scale Image Super-Resolution in Real World (CVPR 2024 · similarity 0.84)
- Occlusion-Embedded Hybrid Transformer for Light Field Super-Resolution (AAAI 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.