Deep Learning for Anisoplanatic Optical Turbulence Mitigation in Long Range Imaging
Abstract
The acquisition of long range images is often affected by atmospheric optical turbulence. Atmospheric turbulence causes degradation in theimages that lowers their utility. The turbulence degradation is caused by changes in temperature and pressure that causes quasi-periodicspatial and temporal blur as well as warping [1,35]. The blur and warp is an effect of variations in the index of refraction along the opticalpath. In the field of astronomy, this has been heavily investigated [4]. For astronomy the field-of-view tends to be narrow which causes thescene to be affected by the atmosphere evenly across the entirety of the image. The uniform effect of this degradation allows one to model the blurring of the entire scene with one linear spatially invariant (LSI) point spread function (PSF) and the warping is only global shifting. This type of imaging scenario with a narrow field-of-view is called isoplanatic [1]. In the case of terrestrial long-range imaging of extended scenes there is a wide field-of-view. This leads to the atmosphere not being uniform over the entire scene, requiring multiple PSFs to be used to successfully model the degradation. This scenario is referred to as anisoplanatic. Turbulence mitigation (TM) in the case of anisoplanatic conditions tends to be a very challenging problem and this is what we address in this thesis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 15, 2020
- Accession Number
- AD1117810
Entities
People
- Matthew A. Hoffmire
Organizations
- University of Dayton