A SIMPLE, PHOTOMETRICALLY ACCURATE ALGORITHM FOR DECONVOLUTION OF OPTICAL-IMAGES

A modified version of the CLEAN deconvolution technique, incorporating a noise model, has been tested on simulated HST data. This straightforward algorithm displays many desirable features for astronomical image reconstruction, as shown by rigorous numerical tests: (1) The photometric scale of the data is preserved accurately. (2) Information on the local noise level due to Poisson statistics in the input data is preserved. (3) It can accommodate a spatially variable point-spread function. Some of these properties are not shared by any widely used deconvolution algorithm. Some considerations of computing speed, and application to ground-based images, are briefly discussed.