November 30, 2013 7:00 pm JST

Network surveillance systems use computing power to spot threats in crowd

TAKANORI OKABE, Nikkei staff writer

TOKYO -- Japanese and U.S. companies are developing network surveillance and monitoring programs that can spot dangers and individual persons in crowds with far greater accuracy and speed than ever before. Much of this is thanks to big data and faster computers.

    Earlier this month, NEC announced that it has developed the world's first crowd behavior analysis technology. Based on the simulated behavioral patterns exhibited by people in emergencies, the system is designed to detect any abnormalities in the behavior of congested public places.

     The technology, NEC says, is intended to prevent crimes and terrorist attacks. The company is currently testing the system, aiming to have it on the market by fiscal 2014.

     The new technology can assess what is happening even in a crowd-packed environment by analyzing the collective behavior of the situation rather than the activities of individuals. It has preinstalled patterns of hundreds of thousands of sample images pertaining to the flow, direction and density of crowds. When it detects deviations to these common behaviors, such as people fleeing or others gathering in unusual clusters, it issues an alert.

     "Until now, our technology was unable to detect patterns in a large crowd," said Toshihiko Hiroaki of NEC Information and Media Processing Laboratories. "But focusing on any signs of change in crowd behavior, we can quickly detect abnormalities."

One among billions

Also this month, Hitachi Kokusai Electric began marketing a new surveillance system that can search and identify a target individual by using an enormous volume of recorded footage from surveillance cameras. The company extracts facial features of individuals -- including profile shape, eye size and the shape of a nose bridge -- and stores them in a database. The system can then compare the features of the person in question against this data. It can complete the search within one second using a database of 36 million faces.

     This search technology could be used to quickly identify terrorists in public places. The company said the system could register faces of up to 7 billion people if 50 servers are linked up.

     "It can be used for public facilities such as airports and railway stations, but it can also be used to search for missing family members in times of disaster. So, this system has a wide range of uses," said Yuichi Onami of Hitachi Kokusai Electric. "We want to sell it to overseas markets."

     With the penetration of Internet-accessible cameras, surveillance and monitoring systems are becoming larger in scale. For example, such a system could consolidate images and footage taken by multiple cameras into a server or other storage infrastructure, and efficiently manage and analyze stored image data.

Olympic moment

According to Tokyo-based Yano Research Institute, the global shipment of network surveillance cameras is forecast to reach 5.75 million units in 2015. In 2012, there were 3.21 million units shipped. The average annual growth rate exceeds over 22% from 2009 to 2015, the research institute says. The strongest demand for such systems mainly in countries across Asia and the Middle East.

    EMC of the U.S., a leading information-technology storage hardware provider, has developed a surveillance system that uses external storage technology. The system can hold data of up to 20 petabytes by linking up as many as 144 storage devices. This is enough to store 13 years of continuous footage from 100 cameras.

     The company says this system could be useful for monitoring large areas, such as an entire town or a major event.

     "A huge number of athletes and tourists will come to the Tokyo Summer Olympics (in 2020) from around the world," said Masaaki Hatori, a manager at EMC Japan. "We want to capture this Olympic-related demand for surveillance systems."