Abstract
Correction of atmospheric turbulence effects on images involves mainly mitigation of distortion (“de-warping”) and removal of image blur. One of the approaches for correcting atmospheric blurring involves the use of deconvolution. The ill-posed nature of the problem and the number of unknowns makes this problem hard to solve. This is why methods like blind deconvolution can be too time-consuming for real-time application. Additionally, an optimal parameter input is also often required (which requires interaction from an operator). Our ultimate goal is to perform an autonomous, software-based turbulence correction in real-time. This requires both very fast point-spread function (PSF) estimation and a deconvolution method. In this work we study new efficient ways to describe and estimate the PSF in anisoplanatic conditions.