Research map / Image generation / Super-resolution
HoloADMM: High-Quality Holographic Complex Field Recovery
Mazen Mel, Paul Springer, Pietro Zanuttigh, Haitao Zhou, Alexander Gatto
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 |
|---|---|---|---|
| replace competing learned reconstruction networks with unrolled ADMM and learned prior (no SR) | HoloADMM · Synthetic Do, phase | PSNR (dB): 23.822 -> 28.092 | Improved |
| use HoloADMM with x4 downsampled inputs versus baselines on full-resolution inputs | HoloADMM · Synthetic Do, phase | SSIM: 0.703 -> 0.791 | Improved |
| increase super-resolution factor from x4 to x8 (ablation) | HoloADMM · Synthetic Do, phase | PSNR (dB): 27.626 -> 20.794 | Hurt |
| use 10 holograms instead of 1 at inference (ablation) | HoloADMM · Do Airplane, phase | PSNR (dB): 24.828 -> 28.205 | Improved |
| replace iterative solvers with learned unrolled network | HoloADMM · Do Airplane, phase | time (s): 990 -> 0.29 | Improved |
| increase number of unrolled ADMM steps (ablation) | HoloADMM · small training subset | training loss | Improved |
| replace unrolled ADMM with a straightforward network mapping hologram stack to latent field (ablation) | HoloADMM · small training subset | training loss | Hurt |
| add image registration of back-propagated holograms (ablation) | HoloADMM · super-resolution | image quality | Improved |
| use a smaller synthetic training dataset than 100k samples (ablation) | HoloADMM · synthetic training | reconstruction quality | Hurt |
| apply competing learned models trained only on synthetic data to real holograms | baselines · real DIHM holograms | reconstruction quality | Hurt |
Closest papers by meaning
- Towards Mixed-State Coded Diffraction Imaging (TPAMI 2022 · similarity 0.85)
- 3PointTM: Faster Measurement of High-Dimensional Transmission Matrices (ECCV 2020 · similarity 0.84)
- V2V3D: View-to-View Denoised 3D Reconstruction for Light Field Microscopy (CVPR 2025 · similarity 0.82)
- Single Exposure Quantitative Phase Imaging with a Conventional Microscope Using Diffusion Models (AAAI 2025 · similarity 0.81)
- OpticalNet: An Optical Imaging Dataset and Benchmark Beyond the Diffraction Limit (CVPR 2025 · similarity 0.81)
- MicroDiffusion: Implicit Representation-Guided Diffusion for 3D Reconstruction from Limited 2D Microscopy Projections (CVPR 2024 · similarity 0.80)
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.
