Research map / 3D vision / Depth and stereo
Light Field Reconstruction via Deep Adaptive Fusion of Hybrid Lenses
Jing Jin, Mantang Guo, Junhui Hou, Hui Liu, Hongkai Xiong
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 5x5 instead of 3x3 LR side views (ablation) | Ours · Simulated hybrid LF (8x) | PSNR/SSIM: 38.65/0.983 -> 39.51/0.986 | Improved |
| use stack of LR side images in SR-Net instead of HybridLF-Net SR-Net (ablation) | Ours SR-Net · Simulated hybrid LF (8x) | PSNR/SSIM: 36.02/0.973 -> 37.17/0.978 | Improved |
| use multi-scale Warp-Net instead of HybridLF-Net Warp-Net (ablation) | Ours Warp-Net · Simulated hybrid LF (8x) | PSNR/SSIM: 33.38/0.959 -> 33.97/0.963 | Improved |
| use multi-scale instead of single-scale Warp-Net (ablation) | Ours Warp-Net · Simulated hybrid LF (8x) | PSNR/SSIM: 33.82/0.961 -> 33.97/0.963 | Improved |
| place HR image at central view instead of top-left corner (ablation) | Ours · Simulated hybrid LF (4x) | PSNR/SSIM: 38.29/0.982 -> 39.91/0.987 | Improved |
| place HR image at central view instead of top-left corner (ablation) | Ours · Simulated hybrid LF (8x) | PSNR/SSIM: 35.07/0.966 -> 37.34/0.979 | Improved |
| add color augmentation of side views in training (ablation) | Ours · Real hybrid data | visual quality | Improved |
| add confidence-based fusion of SR-Net and Warp-Net outputs (ablation) | Ours · Real hybrid data | visual quality | Improved |
| add the extra HR view to the input (M-RDN to M-RDN-H) | M-RDN · Simulated hybrid LF (8x) | PSNR: 28.20 -> 35.13 | Improved |
| reduce side-view resolution from 8x to 16x downsampling (ablation) | Ours · Simulated hybrid LF | PSNR/SSIM: 37.34/0.979 -> 32.20/0.944 | Hurt |
| our method vs HybridLF-Net | Ours · Simulated hybrid LF (8x) | PSNR: 36.22 -> 37.34 | Improved |
| our method vs HybridLF-Net | Ours · Simulated hybrid LF (4x) | PSNR: 39.43 -> 39.91 | Improved |
Closest papers by meaning
- Fast Light Field Reconstruction With Deep Coarse-To-Fine Modeling of Spatial-Angular Clues (ECCV 2018 · similarity 0.91)
- Deep Coarse-to-Fine Dense Light Field Reconstruction With Flexible Sampling and Geometry-Aware Fusion (TPAMI 2020 · similarity 0.90)
- High-Dimensional Dense Residual Convolutional Neural Network for Light Field Reconstruction (TPAMI 2019 · similarity 0.89)
- Deep Spatial-angular Regularization for Compressive Light Field Reconstruction over Coded Apertures (ECCV 2020 · similarity 0.89)
- Deep Spatial-Angular Regularization for Light Field Imaging, Denoising, and Super-Resolution (TPAMI 2021 · similarity 0.89)
- NeLF-Pro: Neural Light Field Probes for Multi-Scale Novel View Synthesis (CVPR 2024 · similarity 0.86)
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.
