Deconvolution and photometry of Faint Object Camera images with point sources.

Abstract

The peculiar characteristics of images obtained with the Hubble Space Telescope have raised important questions concerning the proper method for processing data in which the point spread function is seriously distorted. In particular, there is considerable confusion over whether or not it is useful to deconvolve the images, and to what extent the various algorithms available preserve flux. In an effort to clarify these issues, we have undertaken an empirical investigation of the effect of deconvolution on both point sources and extended sources in FOC images. In this paper, we present the results of experiments with low and high signal-to-noise ratio images of point sources in crowded and uncrowded fields. We use MEMSYS-5 for most of the deconvolutions, although we also comment on the Lucy method. We use both simulated and real FOC images of R136, the group of very massive stars in the central region of 30 Doradus in the Large Magellanic Cloud. Our basic conclusion is that, for fields containing only point sources, deconvolution of typically low signal-to-noise ratio FOC images does not significantly improve the range or accuracy of photometry over that which can be achieved simply by using apertures of an appropriate size on an undeconvolved image. FOC images with significantly higher signal-to-noise ratio may benefit from deconvolution, since the information in the extensive, low surface brightness wings of the PSF could then be properly exploited.