Research map / 3D vision / Depth and stereo
Learning to Capture Light Fields through a Coded Aperture Camera
Yasutaka Inagaki, Yuto Kobayashi, Keita Takahashi, Toshiaki Fujii, Hajime Nagahara
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 the refinement network NR2 (VDSR-style) after the tentative reconstruction network (ablation) | NA+NR1 coded-aperture auto-encoder · Dragon and Bunnies, 5x5, N=2 | PSNR | Improved |
| train NA+NR1+NR2 end-to-end instead of training parts separately then fine-tuning (ablation) | NA+NR1+NR2 · Dragon and Bunnies, 5x5, N=2 | PSNR | No clear effect |
| test with noise larger than the training noise level (ablation) | trained network · Dragon and Bunnies, 5x5, N=2 | PSNR | Hurt |
| increase number of acquired images from 1 to 2 (ablation) | learned coded aperture camera · light field datasets, 5x5 and 8x8 | PSNR | Improved |
| increase number of acquired images from 2 to 4 (ablation) | learned coded aperture camera · light field datasets, 5x5 and 8x8 | PSNR | Improved |
| use fixed random aperture patterns instead of learned patterns (ablation) | learned aperture patterns · 5x5, N=2 | PSNR | No clear effect |
| lower the brightness of the aperture patterns (ablation) | fixed aperture patterns · 5x5, N=2 | PSNR | Hurt |
| constrain the second aperture pattern to be the first rotated by 90 degrees (ablation) | learned aperture patterns · 5x5, N=2 | PSNR | No clear effect |
| learned coded aperture versus Yagi's PCA/NMF compressive sensing | Yagi et al. · Dragon and Bunnies, Medieval2, 5x5 | PSNR | Improved |
| learned coded aperture versus Marwah's sparse dictionary method | Marwah et al. · Dragon and Bunnies, Medieval2, 5x5 | PSNR | Mixed |
Closest papers by meaning
- Acquiring Dynamic Light Fields through Coded Aperture Camera (ECCV 2020 · similarity 0.93)
- Acquiring a Dynamic Light Field Through a Single-Shot Coded Image (CVPR 2022 · similarity 0.92)
- Deep Spatial-angular Regularization for Compressive Light Field Reconstruction over Coded Apertures (ECCV 2020 · similarity 0.91)
- Time-Efficient Light-Field Acquisition Using Coded Aperture and Events (CVPR 2024 · similarity 0.90)
- Deep Spatial-Angular Regularization for Light Field Imaging, Denoising, and Super-Resolution (TPAMI 2021 · similarity 0.89)
- Light Field Intrinsics With a Deep Encoder-Decoder Network (CVPR 2018 · similarity 0.89)
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.
