What do you gain from deconvolution? - Observing faint galaxies with the Hubble Space Telescope Wide Field Camera

We describe experiments with deconvolutions of simulations of deep HST Wide Field Camera images containing faint, compact galaxies to determine under what circumstances there is a quantitative advantage to image deconvolution, and explore whether it is (1) helpful for distinguishing between stars and compact galaxies, or between spiral and elliptical galaxies, and whether it (2) improves the accuracy with which characteristic radii and integrated magnitudes may be determined. The Maximum Entropy and Richardson-Lucy deconvolution algorithms give the same results. For medium and low S/N images, deconvolution does not significantly improve our ability to distinguish between faint stars and compact galaxies, nor between spiral and elliptical galaxies. Measurements from both raw and deconvolved images are biased and must be corrected; it is easier to quantify and remove the biases for cases that have not been deconvolved. We find no benefit from deconvolution for measuring luminosity profiles, but these results are limited to low S/N images of very compact (often undersampled) galaxies.