Discussion by Nicholas Weir

Professor Ripley clearly presents the Bayesian paradigm as it applies to astronomical image reconstruction. The particular prior model he presents for image deconvolution appears to be a powerful one and an interesting alternative to the popular maximum entropy (MaxEnt) prior. Perhaps most important, however, he describes how our notion of restored images may be enlarged to include model spaces far different from the traditional pixel grid we associate with deconvolution. If recent results are any indication, research involving novel image models and their priors will soon provide the most significant breakthroughs in astronomical image restoration since Bayesian methods were first applied to the field nearly two decades ago (see [Pri72]).