Yea, that’s very interesting - makes the djembe ‘easier’ to play. Something like that would be interesting to explore, but perhaps multitouch capable - an ordinary drum set’s capabilities could be expanded to support other gestures performed on the surface of the drum.
This paper is somewhat similar to regular electronic drum sets - I think their approach could be coupled with the experiment you’re doing.
They used the Creative NX Ultra webcam - I haven’t read of anyone using it on the forums, but perhaps that’s also a piece of hardware that could be tried out. 78 degree view angle is nice, and I imagine it would have to have a reasonably good frame rate to do what they were doing. According to Amazon, its max resolution is 640 x 480, but no word on the frame rate. They did mention later in the paper that users said that latency was the biggest problem, and they would drop out of rhythm when they did not actually make a mistake.
Replicating what they did with your drum prototype shouldn’t be very difficult - just use Flash to write a program that divides up the drum in several regions, with each region producing a different sound. They used the ‘height of the hand’ to control the volume, but I imagine that with a FTIR setup, the pressure of the blob could control the volume - front DI [mtmini] isn’t truly pressure sensitive, though.
Cool link...let us know how your drum goes 
If you don’t mind, i’ll put it up on my blog.
