Ha, I'll keep an eye out for your publications. I'm particularly interested at the distance you'll have to go in gathering data, and what will you be able to make out of what is known. I expect that scans aiming for connectome description contain some neuron type data already due to morphological differences in neurons. I don't know what sets of sensors are used for those scans, but maybe getting a broader spectrum could provide clues as to what neuron types occupy which space inside the connectome. SEM can, after all, determine the chemical composition of materials, can't it?. As-is, this seems a pretty neckbreaking undertaking, although I wish you the best of luck.

In other news, there is, luckily, more and more work in this field: http://www.theverge.com/2011/11/16/2565638/mit-neural-connectivity-silicon-synapse

Predictions for silicon-based processors are pretty optimistic as well - Intel aims to achieve 10nm by 2014, and similar date is pushed by nVidia. Past that date we may see some major leaps in available technology (or not), and development of multi-processor computation algorithms is finally gaining momentum since Von Neumann's Big Mistake.

Maybe the Kurzweil's 2025 date for brain emulation is a bit overoptimistic, but I don't expect that to take much longer. I do think that the first dozen of successful neural structure emulations will become a significant breakthrough, and we'll see a rapid expansion similar to that in genetic sciences not so long ago.