A patent recently filed by Google for an interactive TV service is chock full of new details, but the most interesting is this one: "an image capture device (e.g., digital camera, video recorder, etc.) can be used to measure how many viewers are watching or listening to a broadcast."
According to the patent, the main point of the system is to identify audio within a TV broadcast and compare it to the appropriate reference material for identification. After that, it aggregates "personalised information related to the media broadcast."What sort of personalised info? On one hand, Google wants to layer in complementary websites and other data—I hope this means the IMDB tab I've been waiting for, but it also refers to the old "buy what the hot actress on Letterman is wearing" function. On the other hand, Google is really trying to turn your TV watching into a social experience that would be a lot smoother than logging into a computer. The goal is to share a viewing and chatting experience simultaneously, but this doesn't seem to account for the fact that DVRs have made realtime TV watching an antiquated concept.
More importantly to us, but way down in the filing, another form of analysis besides audio evaluation is discussed, that of using a still or video camera for "pattern matching":
Various known pattern-matching algorithms can be applied to an image or a sequence of images to determine the number of viewers present in a broadcast environment during a particular broadcast. The images and or data derived from the images can be used in combination with audio descriptors to gather personalised information for a user, compute popularity ratings, or for any other purpose.