When the particulate filter is 'full' and the burn-off process occurs doesn't this release the same particulate material, just converted into an even finer form, and in a different location?

Not in the sense that you mean. The ideal oxidation reaction of the regenerative burn-off converts the soot particles (primarily carbon) to CO2 and H20—think of it as gasifying the soot.

In practice, given the different attributes of soot, and the regenerative process, there will likely be CO, HC and other emissions attendant.

So, yes, you’re outputting more CO2, CO and a level of ride-along contaminants, but no, you’re not outputting more PM.

Checking for and managing secondary emissions is a big part of the development and testing process of the DPF.

The purpose of the filter, though, is not to eliminate all emissions—it’s specific to dealing with the soot particles that have their own toxic attributes based on size, etc.

Something that has bothered me for a long time is the issue of particle size when discussing PM filters. While it is incontrovertible that the filters reduce PM output by mass, do they reduce the number of fine particles? This is of course the real objective, as particles on the order of 100nm and smaller are the biggest health concern. If you eliminate PM2.5 and PM10 at the cost of higher levels of particles with diameters on the nanometer scale, you have actually made the health situation worse.

DPFs have been shown to be about equally effective across the entire size distribution range. As a matter of fact, many studies have shown that the exhaust of a DPF-equipped engine has (numerically) fewer aerosols than the intake air. See, for example, slide 10 in this presentation:

http://www.osti.gov/fcvt/deer2003/storeypresentation.pdf .