<title>Application of a deconvolution technique to enforce a-priori object constraints to 2-D IR speckle data</title>

Abstract

ABSTRACT Recently, an iterative deconvolution algorithm has been proposed to enforce a priori object constraintson speckle interferometric data. This paper demonstrates its application to both simulated and real 2-D infrared speckle data and compares its performance to that of an iterative transform algorithm. This iterative deconvolution algorithm differs from the iterative transform algorithm in that the object's Fouriermodulus, as well as the Fourier phases, is also constrained by the a priori information.1. INTRODUCTION The application of two-dimensional array detectors at near-infrared wavelengths has revolutionized infrared astronomy' . Such arrays are also applicable to infrared speckle interferometry and there are currently a number of such cameras, some of which have been in use for a number of years. One of these is the Kitt Peak National Obser-vatory (KPNO) infrared imager2, 3, as modified for use as a speckle camera4. This instrument has been routinely usedto obtain IR speckle data which is currently being reduced by using Power Spectrum analysis to obtain the Fourieramplitudes and an application of Knox-Thompson algorithm5 to obtain the Fourier phases. The Knox-Thompsonalgorithm, like the bispectrum6 and unlike the power spectrum, is an image correlation technique which retains theobject's Fourier phases, albeit scrambled in the form of phase differences. Recently alternative reduction procedureshave also been used for the data reduction and these are discussed elsewhere in this volume7'8.For the analysis presented in this paper the data was reduced as follows. Firstly the raw data frames are calibratedfor detector bias, bad pixels and fiatfield when necessary (a more detailed discussion of this procedure is given byChristou ei at.5). The ensemble average power spectra and Knox-Thompson cross-spectra are then computed for theobject data, the point source reference data and nearby data. These sky measurements allow the calibrationof the measured Fourier spectra for additive noise due to thermal emission from the sky and telescope and readoutnoise from the detector.Representing apecklegram (individual short-exposure image) as the convolution of the object distribution O(r) withan instantaneous combined telescope-atmosphere point spread function, P(1), i.e.