The field of image reconstruction is awash with various techniques from the very simple (Wiener filters) to the more complicated (Pixon maps). MDC's technology encompasses most of these as they are all, to a greater or lesser extent, approximations to the real problem of maximising the Bayesian posterior. In making these approximations, generality is necessarily lost. MDC's library of methods allows full flexibility to be retained in respect of choice of noise model, signal model and maximisation technique. A core piece of our IP allows automatic choice between these possible models.

In particular, the Pixon method is an attempt to perform a restricted version of MDC's simplest multi level image decomposition, with the aim of speeding up the maximisation procedure. MDC's superior maximiser is efficient enough that such restrictions are not required, so the full Bayesian posterior can be used as the objective function.

We own a filed patent protecting our novel use of selection function to determine the best reconstruction regime. We have a second patent in the drafting stage.