Google Image Swirl organizes image search results into groups and sub-groups, based on their visual and semantic similarity and presents them in an intuitive exploratory interface.
There's this branch in computer science and statistics for vision research. Normally, if you ever hear about it in the news it's in the context of spotting terrorists in security tapes or facial recognition checkpoints (you know, like what they have in movies in front of giant steel doors). That is of course not the only application.
Google (and many others) has been playing around with this stuff for a while. Most recently, they released Google Image Swirl in their labs section, which utilizes computer vision to find similar images." [Flowing Data]