Demo: Seamless Surveillance

Sarnoff shows how to turn the feeds from many surveillance cameras into a unified 3-D scene.

Video cameras are proliferating; they’re everywhere at airports, urban centers, and government buildings. But how do you tackle the tedious job of actually watching all these boring, narrowly focused video feeds? Sarnoff, the former RCA Labs now owned by SRI International, is building a solution-a system that combines video from many cameras into a 3-D model of an area. “Instead of watching the world through a soda straw, this is essentially taking video and putting it into context,” says Rakesh Kumar, a computer scientist who developed the technology as director of Sarnoff’s 14-member media vision lab. The result is called Video Flashlight: it’s like playing a video game, except the scene is of real events in real time; grab a joystick and you can swoop down hallways and fly around buildings, immersing yourself in a scene. The technology has a rich legacy: RCA was a pioneer in television technology, and its laboratory was a key source of World War IIera electronics innovations. Today, Video Flashlight is getting tested as a security tool at government buildings. At Sarnoff’s labs in Princeton, NJ, Kumar showed TR senior editor David Talbot the latest in surveillance.