Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform
Abstract
Fast and accurate simulation of imaging through atmospheric turbulence is essential for developing turbulence mitigation algorithms. Recognizing the limitations of previous approaches, we introduce a new concept known as the phase-to-space (P2S) transform to significantly speed up the simulation. P2S is built upon three ideas: (1) reformulating the spatially varying convolution as a set of invariant convolutions with basis functions, (2) learning the basis function via the known turbulence statistics models, (3) implementing the P2S transform via a light-weight network that directly converts the phase representation to spatial representation. The new simulator offers 300x - 1000x speed up compared to the mainstream split-step simulators while preserving the essential turbulence statistics.
Cite
Text
Mao et al. "Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform." International Conference on Computer Vision, 2021. doi:10.1109/ICCV48922.2021.01449Markdown
[Mao et al. "Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform." International Conference on Computer Vision, 2021.](https://mlanthology.org/iccv/2021/mao2021iccv-accelerating/) doi:10.1109/ICCV48922.2021.01449BibTeX
@inproceedings{mao2021iccv-accelerating,
title = {{Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform}},
author = {Mao, Zhiyuan and Chimitt, Nicholas and Chan, Stanley H.},
booktitle = {International Conference on Computer Vision},
year = {2021},
pages = {14759-14768},
doi = {10.1109/ICCV48922.2021.01449},
url = {https://mlanthology.org/iccv/2021/mao2021iccv-accelerating/}
}