Research Highlights

Security Cameras that Think for Themselves (2010.7.26)

2010-11-15 1,099

In the D. J. Caruso film Eagle Eye (2008), a super-computer equipped with artificial intelligence controls the society by monitoring everything with CCTVs located all over the world. A recent development suggests that the movie may possibly become a reality; a surveillance camera that can judge a situation by itself and prevent terrorism or theft has been invented.

A research team led by Professor Daejin Kim of the Department of Computer Science and Technology has developed a technology which automatically detects unattended items as well as burglaries in public places such as subway stations and airports. It detects a foreground region that was not detected conventionally, singles out the region that is still without any sort of movement, and then figures out whether the item inside the region is left behind or taken away.

To increase its capacity, the camera operates through a couple of steps. At first, it scans for all regions that are stationary, and then it uses a coarse-to-fine strategy to finalize the detected stationary region. The factors that help the process of this strategy include a region concealed by an animated object, color difference between the region and object, and other indications of movement.

The new technology is expected to help prevent various crimes, from minor offenses such as illegal trash dumping to major cases like terrorist bombing. Also, as it can immediately capture any incident of theft, it can be useful in exhibitions or displays, and military use is also possible in operations such as detection of mines and bombs, or preemption of ambush by the enemy.

Professor Daejin Kim stated that the team’s recent discovery has increased the possibility of success by 20 percent compared to the American technology in commercial use today. He furthermore noted that the technology is more than 15 percent less prone to error. Also, he pointed out that it is more flexible in the sense that it can be mounted on moving as well as stationary security cameras, whereas the conventional method can only be used on stationary ones. Moving cameras include active cameras or mobile surveillance robots.

Professor Takeo Kanade from Carnegie Mellon University, USA, was part of this World Class University joint research team and also Samsung Techwin participated as an industrial block. Consequently, an arrangement is being made for an application to Samsung Techwin futuristic surveillance cameras that will be released on the first half of 2011, and utilized in various fields that need intelligent detection.