There exist several image-enhancement algorithms and tasks associated with imaging through turbulence that depend on defining the quality of an image. Examples include: “lucky imaging”, choosing the width of the inverse filter for image reconstruction, or stopping iterative deconvolution. We collected a number of image quality metrics found in the literature. Particularly interesting are the blind, “no-reference” metrics. We discuss ways of evaluating the usefulness of these metrics, even when a fully objective comparison is impossible because of the lack of a reference image. Metrics are tested on simulated and real data. Field data comes from experiments performed by the NATO SET 165 research group over a 7 km distance in Dayton, Ohio.

Date Published: 31 May 2013
PDF: 12 pages
Proc. SPIE 8720, Photonic Applications for Aerospace, Commercial, and Harsh Environments IV, 872006 (31 May 2013); doi: 10.1117/12.2015710

Published in SPIE Proceedings Vol. 8720:
Photonic Applications for Aerospace, Commercial, and Harsh Environments IV
Alex A. Kazemi; Bernard C. Kress; Simon Thibault, Editor(s)