Stephen Lewis's Processing file test.pde tracks multiple blobs and turning them into a tone melody. Clicking the mouse on any pixel re-assigns that pixel's color to be the tracking color, but it starts on black, so it can follow shadows on the backlit table top. The table is actually a piece of mylar inside a plexiglass holder suspended above a camera in a garbage can. All materials available at Staples, including the garbage can, mylar (or tissue paper) and plexi. You can also buy a webcam at Staples.
His program uses the ESS library which makes audible tones. ESS is at http://www.tree-axis.com/Ess/, but also listed on the processing help file under "libraries". The JMyron vision library is at http://webcamxtra.sourceforge.net/, but also listed directly in Processing's reference help file. Expect it to run at about 5 frames per second.
There's also a brightness tracking library right in Processing's examples menu, which uses a brute force pixel-examiner algorithm to find the brightest pixel (no centroid or extents or anything about the bright-pixel being in a blob, so it's a bit random and jittery). Expect that to run at around 30 frames per second.