Research map / Image generation / Super-resolution
Light Field Super-Resolution using a Low-Rank Prior and Deep Convolutional Neural Networks
Reuben A. Farrugia, Christine M. Guillemot
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 iterative back-projection post-processing (ablation) | LR-LFSR · Bee 2 (INRIA) x2 SR | PSNR: 33.4915 -> 35.0306 | Improved |
| add iterative back-projection post-processing (ablation) | LF-SRCNN · Bee 2 (INRIA) x2 SR | PSNR: 33.8268 -> 34.7257 | Improved |
| add iterative back-projection post-processing (ablation) | LR-LFSR · Chess (STANFORD) x3 SR | PSNR: 30.1485 -> 31.4388 | Improved |
| IBP applied to both methods widens the gap between LR-LFSR and LF-SRCNN (ablation) | LR-LFSR vs LF-SRCNN · INRIA and Stanford light fields | PSNR: 0.23 dB and 0.12 dB without IBP, roughly doubles with IBP (0.41 dB and 0.31 dB) | Improved |
| LR-LFSR vs second-best LF-SRCNN, no IBP | LR-LFSR · INRIA and Stanford light fields | PSNR: 0.2 dB gain | Improved |
| LR-LFSR with IBP vs LF-SRCNN with IBP | LR-LFSR-IBP · INRIA and Stanford light fields | PSNR: +0.41 dB at x2, +0.31 dB at x3 | Improved |
| per-view single-image SR (VDSR) vs bicubic | VDSR · INRIA and Stanford light fields | PSNR: +0.33 dB over bicubic | Improved |
| add block-matching alignment to PCA+RR | PCA+RR · INRIA and Stanford light fields | PSNR: 2.52 dB and 1.76 dB over bicubic | Improved |
| LF-SRCNN vs BM+PCA+RR | LF-SRCNN · INRIA and Stanford light fields | PSNR: marginal gain 0.05 dB | No clear effect |
| no disparity compensation (PCA+RR) on large-disparity Stanford light fields | PCA+RR · Bracelet (STANFORD) x2 SR | PSNR: 23.8183 vs bicubic 26.4662 | Hurt |
| LR-LFSR vs PCA+RR on small-disparity INRIA | LR-LFSR · Bee 2 (INRIA) x2 SR | PSNR: 33.4915 vs PCA+RR 34.1579 | Hurt |
Closest papers by meaning
- High-Dimensional Dense Residual Convolutional Neural Network for Light Field Reconstruction (TPAMI 2019 · similarity 0.94)
- Residual Networks for Light Field Image Super-Resolution (CVPR 2019 · similarity 0.94)
- Light Field Super-Resolution With Zero-Shot Learning (CVPR 2021 · similarity 0.92)
- Learning Light Field Angular Super-Resolution via a Geometry-Aware Network (AAAI 2020 · similarity 0.92)
- Detail-Preserving Transformer for Light Field Image Super-resolution (AAAI 2022 · similarity 0.91)
- Disentangling Light Fields for Super-Resolution and Disparity Estimation (TPAMI 2022 · similarity 0.91)
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.
