Some aspects of statistical image modelling and restoration
In: Phystat05, 12-15 September 2005, Oxford, Englamd.
A review is provided of some ways in which statistical
ideas have influenced research into image
analysis, in particular the problems involved in making inferences about the
true scene and any parameters in the underlying model. The emphasis will
be on the application of general statistical paradigms such as maximum
likelihood, implemented using tools such as the EM algorithm,
and Bayes' Theorem. The resulting procedures can be regarded as particular
recipes for regularisation or deconvolution, according to the context.