Research map / Architectures and efficiency / CNN design
Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform
Zhiyuan Mao, Nicholas Chimitt, Stanley H. Chan
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 |
|---|---|---|---|
| train the U-Net on data from the P2S simulator instead of the Lau et al. simulator | U-Net turbulence reconstruction · Test D/r0=1.5 | PSNR (dB): 26.59dB -> 27.18dB | Improved |
| train the U-Net on data from the P2S simulator instead of the Lau et al. simulator | U-Net turbulence reconstruction · Test D/r0=3.0 | PSNR (dB): 26.11dB -> 26.98dB | Improved |
| train the U-Net on data from the P2S simulator instead of the Lau et al. simulator | U-Net turbulence reconstruction · Test D/r0=4.5 | PSNR (dB): 25.40dB -> 26.01dB | Improved |
| use P2S U-Net instead of deterministic Mao et al. method at strong turbulence | turbulence reconstruction · Test D/r0=4.5 | PSNR (dB): 25.85dB -> 26.01dB | Improved |
| use P2S U-Net instead of deterministic Mao et al. method at weak and moderate turbulence | turbulence reconstruction · Test D/r0=1.5 and 3.0 | PSNR (dB): 27.33dB -> 27.18dB (D/r0 1.5); 27.04dB -> 26.98dB (D/r0 3.0) | No clear effect |
| replace split-step simulation by P2S simulation | turbulence simulator · 256x256 frame, CPU | run time (s): 119.63 -> 0.35 | Improved |
| run P2S on GPU rather than CPU | P2S simulator · 256x256 frame | run time (s): 0.35 -> 0.026 | Improved |
| replace Chimitt-Chan collapsed model by P2S simulation | turbulence simulator · 256x256 frame, CPU | run time (s): 5.88 -> 0.35 | Improved |
| replace Lau et al. subsampling by P2S simulation | turbulence simulator · 256x256 frame, CPU | run time (s): 3.13 -> 0.35 | Improved |
| use one 525nm PSF for all RGB channels instead of three wavelength-specific PSFs (ablation) | color turbulence simulation · qualitative color comparison | visual difference | No clear effect |
| interpolate Zernike coefficients (phase domain) instead of interpolating PSFs spatially (ablation) | grid interpolation in simulator · qualitative PSF example | PSF fidelity (lucky effect) | Improved |
Closest papers by meaning
- Learning Phase Distortion with Selective State Space Models for Video Turbulence Mitigation (CVPR 2025 · similarity 0.89)
- Spatio-Temporal Turbulence Mitigation: A Translational Perspective (CVPR 2024 · similarity 0.85)
- NB-GTR: Narrow-Band Guided Turbulence Removal (CVPR 2024 · similarity 0.84)
- Temporally Consistent Atmospheric Turbulence Mitigation with Neural Representations (NeurIPS 2024 · similarity 0.84)
- PlaNet: Learning to Mitigate Atmospheric Turbulence in Planetary Images (AAAI 2025 · similarity 0.83)
- Diffeomorphic Template Registration for Atmospheric Turbulence Mitigation (CVPR 2024 · similarity 0.83)
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.
