Accelerating Atmospheric Turbulence Simulation via Learned Phase-to-Space Transform
This work addresses the need for faster turbulence simulation to aid algorithm development in imaging applications, representing a novel method for a known bottleneck.
The paper tackles the problem of slow atmospheric turbulence simulation for developing mitigation algorithms by introducing a phase-to-space (P2S) transform, achieving a 300x to 1000x speedup compared to mainstream split-step simulators while preserving essential turbulence statistics.
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.