Back to Volume
Paper: Implementation of the Richardson-Lucy Algorithm in STSDAS
Volume: 61, Astronomical Data Analysis Software and Systems III
Page: 296
Authors: Stobie, E. B.; Hanisch, R. J.; White, R. L.
Abstract: The Richardson-Lucy algorithm (Richardson 1972; Lucy 1974) is a widely used technique for restoring HST images. This method has a number of characteristics that make it well-suited to HST data. The Richardson-Lucy iteration converges to the maximum likelihood solution for Poisson statistics in the data (Shepp & Vardi 1982), which is appropriate for optical data with noise from counting statistics. This method forces the restored image to be non-negative and conserves flux both globally and locally at each iteration. The restored images are robust against small errors in the point-spread function (PSF). Typical implementations of the Richardson-Lucy algorithm require a manageable amount of computer resources (time and memory) representing a reasonable compromise between quick (but sometimes unsatisfactory) methods such as Weiner filtering and the much slower (although sometimes superior) maximum entropy method. The Richardson-Lucy algorithm was implemented as the lucy task in the STSDAS package in late 1990. Two major enhancements to the task have been implemented during 1993. The first enhancement is the masking of bad pixels which allows both the removal of bad data and a better treatment of image boundaries. It also eliminates the problem of large positive offsets to the sky brightness that was sometimes seen when some values of the original data were very negative. Next is the implementation of the accelerated iteration algorithm (Hook & Lucy 1992) which in some cases may reduce the processing time by an order of magnitude. These enhancements will be discussed in detail with examples. Also, future enhancements planned for the lucy task will be discussed.
Back to Volume