English

Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform

Image and Video Processing 2021-08-24 v2 Computer Vision and Pattern Recognition Fluid Dynamics

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 build 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 convert 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.

Keywords

Cite

@article{arxiv.2107.11627,
  title  = {Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform},
  author = {Zhiyuan Mao and Nicholas Chimitt and Stanley H. Chan},
  journal= {arXiv preprint arXiv:2107.11627},
  year   = {2021}
}

Comments

The paper will be published at the ICCV 2021

R2 v1 2026-06-24T04:29:19.145Z