Researcher developing video technology for surveillance

COLUMBUS – An Ohio State University professor is developing an advanced video surveillance system that uses computers equipped with video cameras to not only detect the presence of people, but also to identify their activities.

The research has broad implications for national security as well as search and rescue, border patrol, law enforcement, and many other types of military applications.

James Davis, assistant professor of computer and information science, is developing the system that combines a thermal camera with computer learning methods, enabling the computer to perform the kind of visual recognition that seems effortless for humans.

“We are building an intelligent video surveillance system that can reliably detect the presence of people, track their movements, and recognize their activities and ultimately their intent,” Davis said.

“We use thermal cameras to locate the people, and use statistical methods to classify the activity of the person into different categories such as walking, running, standing, and throwing. We are also interested in more qualitative aspects of the action, like determining if the person is a child or adult, moving leisurely or in a hurry, or carrying light or heavy packages.”

The system uses a thermal camera to measure the amount of heat in a scene. Thermal cameras are already widely used by law enforcement and military agencies, especially to locate people in the dark.

“Thermal images provide a unique signature of humans so they stand out as a bright white image against a dark background,” Davis said.

The camera also picks up residual heat. “If I place my hand on the wall for a short time and then remove it, my handprint sits on the wall,” he explains. “The thermal cameras are able to pick this up where standard video cameras can’t see a thing.

“The computer looks for the white hot spots in the thermal video to locate a potential person. Then it examines the shape and motion of the region and matches it to a database of activity patterns.”

Davis trains the computer by giving it multiple examples of each activity. The machine learns what key features are needed to identify people and recognize their actions.

Davis has worked on human activity analysis with computers for the last 10 years. His research has received funding from the National Science Foundation and the U.S. Army Night Vision Lab.

Davis expects to demonstrate the technology within the next two or three years. “There are automatic systems on the market today that you can purchase,” Davis said. “But we’re expecting these advanced systems to be more widely deployed within the next five to 10 years.”

Davis’ work in investigating computer vision methods was recognized last month by the National Science Foundation with the prestigious NSF Faculty Early Career Development (CAREER) Program award. The awards, which are highly competitive, are given only to the nation’s most outstanding junior researchers. Davis will receive $500,000 over the next five years to continue his research in this area.