I heard Leo Guibas talk the other day. It was refreshing a hear a talk that was about a research agenda, one with deep math and algorithmics, rather than an isolated result. He spoke for example about discovering the structure of given set of geometric objects, leading to a nice definition of a "bar code" that captured betti numbers and other topological properties; lot of the talk described his work with 3D shapes and objects and included some nifty demos (one on finding symmetry within the Coliseum was terrific). Must go, read these papers.
ps: As an aside, 2D barcodes encode more information than the standard 1D barcodes in products and are slowly finding fun applications.