GTC 2020 S21270
Presenters: Michael Werth,The Boeing Company; Kevin Roe,Maui High Performance Computing Center
High-resolution imaging of objects in space from a ground-based observatory is achievable with a sufficiently large aperture, but atmospheric turbulence causes significant degradation. Computationally expensive algorithms can mitigate the blurring effects of turbulence, and these algorithms have only recently begun to leave the domain of CPU-bound computation. We’ll describe space domain awareness, the imaging-through-turbulence problem, and algorithms that attempt to solve it. We’ll also describe Likelihood-based Uncertainty Constrained Iterative Deconvolution (LUCID), a new multi-frame blind deconvolution implementation that uses CUDA to extract high-resolution images of low-Earth-orbit (LEO) satellites from a series of short-exposure observations.
Watch this session
Join in the conversation below.