Smart Software Gives Surveillance Eyes a Brain

Source: University of Rochester News (02/12/04)

Abstract: Computer science laboratory researchers at the University of Rochester have been able to outfit surveillance cameras with a rudimentary software "brain" developed by associate professor of computer science Randal Nelson. The software enables the cameras to look for specific objects, such as a firearm in an airport, or a missing piece of lab equipment, which removes some of the concerns that it is a "Big Brother" tool for invasive surveillance. The software looks for changes within a black-and-white video image, and highlights those changes as it tries to ascertain whether they are caused by the introduction of a new object or the removal of a object that was previously in the scene. The software then takes stock of all the object's colors so that it can zoom in on the object at the operator's request. So that the software can identify specific objects on sight, it is trained to recognize them by studying numerous photos taken from different angles; in this way, a new object will be matched to the software's object image database.

"If we can get intelligent machines to stand in for people in observation tasks, we can achieve knowledge about our environment that would otherwise be unaffordable," Nelson explains. The technology has already been licensed to PL E-Communications, which intends to apply it as a camera control technology for security purposes. "We're hoping to make this technology do things that were long thought impossible--making things more secure without the need to have a human operator on hand every second," states PL E-Communications CEO Paul Simpson. The technology could also be employed by unmanned reconnaissance aircraft to monitor terrain for signs of movement for prolonged periods. www.rochester.edu/news/show.php?id=1698

Descriptors:
Camera technology. Computer software. Electronic surveillance.