Posterior sampling of scientific images

Scientific image processing involves a variety of problems including image modelling, reconstruction, and synthesis. We are collaborating on an imaging problem in porous media, studied in-situ in an imaging MRI in which it is imperative to infer aspects of the porous sample at scales unresolved by the MRI. In this paper we develop an MCMC approach to resolution enhancement, where a low-resolution measurement is fused with a statistical model derived from a high-resolution image. Our approach is different from registration/super-resolution methods, in that the high and low resolution images are treated only as being governed by the same spatial statistics, rather than actually representing the same identical sample.