Face-recognition software spots mugs more accurately when multiple photos of the same person are averaged into a single composite image than when they are presented as individual photos, according to a new study published in Science. The trick works, say researchers Rob Jenkins [left] and A. Mike Burton of the University of Glasgow in Scotland, because it evens out differences in lighting and pose that can make two photos of the same person look much more dissimilar than two shots of different people. Jenkins and Burton tested the procedure on a face-recognition algorithm on a genealogy Web site. The algorithm's success rate improved from 54 percent with individual photos such as Jenkins's passport photo (left of his face in the image) to 100 percent with composites (right of his face). The researchers say improved face recognition could benefit Homeland Security analysts and police trolling for their most wanted. On a lighter note, it also sounds like a killer Facebook app.