Abstract

Most existing techniques to measure wavefront aberrations rely on single or double integrations of local tilt or curvature. These techniques forfeit accuracy when applied to large apertures, when the number of subapertures needed to span the pupil becomes great. Since a natural star must be used to sense overall tilt, it is useful to consider direct interference methods which exploit the spatial coherence of starlight to make large scale phase measurements as well. Phase information is contained directly in the interference speckles of the far field image. Recently Angel et. al.¹ and Lloyd-Hart et. al.² have demonstrated the use of artificial neural nets to analyze speckle images for phase. Here we present the theory behind an approach to reconstruct pistons in real time using data obtained by Fourier transforming the image-plane interference pattern. The underlying idea behind the approach draws on methods developed for long-baseline interferometry³,⁴,⁵.

© 1992 Optical Society of America