The technique of phase diversity was proposed for estimating telescope aberrations for an unknown extended object. The original version of phase diversity requires extensive processing due to a nonlinear optimization algorithm which is prohibitive in a real-time system. Therefore, neural networks were explored as an alternative solution of the problem and this paper will show the modification of the traditional phase diversity method to employ neural networks to estimate aberrations of point source and extended scene data.Simulations indicated aberrations could be estimated to an average error of 0.02 waves RMS.
. INTRODUCTIONThe astronomical community as well as the U.S. Air Force has been interested in larger and larger telescopes due to their greater light gathering capabilities and higher resolutions. One way to achieve an effectively larger telescope is to use multiaperture telescope design. Here the telescope is comprised of smaller multiple telescopes. When the telescopes are properly aligned then the performance is that of a larger single telescope with a mask [1,2J . A scheme must be developed to estimate the misalignments (in this paper, we dealt with relative piston only) and make the appropriate corrections.Techniques have been developed and used for estimating these misalignments for both segmented and multiaperture systems [3][4][5][6] . In general, these techniques are (1) hardware intensive, which poses a problem for space-based telescopes, and (2) they take measurements at the edge of the aperture where the optical performance is the poorest [6] Sandier [7] and Angel [8) have used a modified phase diversity technique along with neural networks to estimate either the atmospheric distortions for a single telescope or the relative misalignments between multiple telescopes. However, their techniques are valid only for point sources. Gonsalves [9,10] and Paxman [11,12] have proposed the method of phase diversity as a technique to measure the misalignments when viewing an extended object. Two images are taken: one at a focused position and the other at a defocused position. Through a nonlinear optimization technique, the misalignments or aberrations are estimated from the two images.In a practical implementation, this method greatly reduces the hardware requirements at a cost of processing time.Therefore, neural networks were investigated here as a replacement for 410/ SPIE Vol. 1982 Photoelectronic Detection and Imaging '93 0-8194-1 229-5/93/$4.OO Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Over the last several years the practical implementation of adaptive optics to compensate for the atmospheric distortions in large telescopes has become a reality. Of the elements that must be considered in the design of an adaptive optics system, the expected atmospheric turbulence is one of the most important. The usual method for estimating these criteria is to use standard atmospheric models or site specific adaptations of these models. An implicit assumption in these models is that the atmosphere can be treated as an isotropic mass and that the index of refraction variations follow Kolmogorov theory.An analysis of two of these features, smoothed data vice a set of individual turbulence profiles and the influence of a partially non-Kolmogorov atmosphere, was performed for a large adaptive optics system. The results show that performance expectations can vary significantly. Smoothed data tends to over estimate atmospheric effects up to 50%. Non-Kolmogorov effects are less significant introducing differences on the order of 10% for zenith observations. The conclusion is that the designer must pay careful attention to the atmospheric model and the method in which it is employed. The use of multiple phase screens created directly from sonde data are recommended.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.