Perhaps I'm missing something. If there's a database of 'augmented' objects, the tablet knows its position and orientation in this augmented space, and the tablet has access to that database, then the object the tablet is pointing towards can be found using only an algorithm and no additional hardware. This is true weather or not there are multiple objects in front of the tablet.

For example ray-tracers use a such an algorithm to determine which surface a cast ray intersects with. I suspect something similar could be done with more simple targets (eg. a large cuboid for a building), and limiting the intersect-able objects based on their relative position.

I may be misinterpreting the discussion, though. I admit, I've only read the last two posts!

YARR!
}:^)~
AugmentedCorrupt