Smart video gains ground
New applications put more power in the hands of users
In the first hours of a double homicide investigation in Cincinnati, police discovered that one of the victim’s car was missing.
After broadcasting a description of the vehicle and its license plate number, an officer in a patrol car thought he might have seen the wanted vehicle.
Traditionally, the officer would alert a supervisor about the possible sighting, said Heather Whitton, project manager of the license plate recognition program at the Cincinnati Police Department’s information technology division.
"The officer would tell the supervisor what the vehicle looked like and where he thought he saw it,” Whitton said.
But in this case, the officer who spotted the missing car was driving one of the city’s eight patrol cars outfitted with automated license plate recognition technology from Federal Signal PIPS. The system includes three cameras mounted on the patrol car’s light bar, and an on board computer collects and interprets license plate numbers.
The officer immediately downloaded the information and fed it into the department’s main server. He conducted a search for the license plate number and a picture of the wanted vehicle appeared.
The system also associates Global Positioning System information with each picture, so officers can see where the photo was taken.
The department shared the photo of the wanted car and its location with other officers, who found the car within hours, Whitton said. Two individuals provided information that led to the arrest of a suspect in the homicide.
Besides playing a significant role in solving the crime, the technology saved the department money by reducing the man-hours needed to investigate the crime, said Cindy Combs, assistant chief of the Cincinnati Police Department.
Although smart-video technology has been around for years, government agencies at all levels are just now beginning to adopt the technology, said Dan Rayburn, an analyst at Frost and Sullivan Inc.
“Agencies are pursuing everything from webcasting meetings to public security applications like border control,” Rayburn said. “The systems have gotten more capable and interactive, so it is no longer about one-way communication.”
Government customers like the technology because, for example, it can help them avoid asking a security guard to stare at a bank of screens, Rayburn said. That method is inefficient and ineffective.
The evolution of related technologies is a catalyst for smart-video applications, he said. For example, police SWAT teams have handheld devices that can receive video, giving them better situational awareness during an incident, he added.
“That wasn’t happening five years ago,” Rayburn said.
One of the biggest advancements in recent years is that cameras now contain the smart-video technology rather than sending video feeds to servers, said Ed Troha, director of global marketing at ObjectVideo Inc.
For example, security systems on U.S. borders have the new technology with video analytics built into cameras, he said.
ObjectVideo has deployed its technology at more than 700 locations along the northern and southern U.S. borders, Troha said.
Most of the locations are entry points, such as roads and railways, which do not have around-the-clock monitoring from border agents. If someone illegally crosses the border at those locations, the technology immediately alerts Customs and Border Protection officials.
“The technology is capable of actively analyzing any scene; that is, any information within the camera’s view,” Troha said. “The software analyzes every pixel within that scene.”
So if a camera is focused on a runway, the technology recognizes when a plane lands, which probably wouldn’t be a security violation. But if the camera detects a human waiting on the tarmac, it would send an alert to security officials.
“The software is capable of distinguishing between a vehicle and a human,” Troha said.
ObjectVideo used to sell the technology primarily as server-based software. The company now sells the technology to manufacturers, which embeds it into chips and devices. Troha said the company changed its strategy because the server model was expensive to buy and support.
The device model is more affordable, and organizations can more easily expand the use of devices, he said.
The types of applications for which government agencies use video have not evolved much, but demand has increased, said Dan O’Neill, president and chief executive officer of Applied Risk Management LLC.
In addition to an increasing number of cameras, security staffs are performing more duties, such as answering phones, dispatching police and responding to e-mail.
“They’re doing lots of things, so something could be happening and they don’t know until after the fact,” O’Neill said.
Even if someone is monitoring video screens, studies show that becomes ineffective after 12 to 30 minutes. More persistent surveillance is needed to protect high-risks targets, experts say.
For example, New Hampshire uses smart-video technology to monitor bridges. The system looks for large trucks that stop on the bridge for any reason. When a truck stops, the system alerts a command center, O’Neill said.
The same technology also monitors the water beneath bridges to ensure that vessels do not enter restricted areas.
Older cameras without video analytics are still valuable, and agencies should assess how to best use them, O’Neill said. They should perform a risk analysis to determine the level of security they need and how many smart cameras they need, he added.
One customer, for example, did not want large vehicles stopping in front of its building for more than a minute. An assessment found that in all cases the goal was not achieved, O’Neill said.
“Because they had someone sitting in the command center, and when a vehicle would pull up, they’d be occupied with other things, and they were missing that 100 percent of time,” he said.
“We sold them one video camera with video analytics — it cost about $8,000 to do it — and the customer went from 100 percent failure to 100 percent success,” he said.
Smart-video technology works well for any task in which employees are asked to multitask, said Craig Cantrell, vice president and general manager at PIPS, the provider of the Cincinnati Police Department’s video technology.
In Cincinnati, an officer would average 200 to 250 log-ins of license plate numbers during an entire shift, Whitton said. But with license plate-recognition technology, an officer can collect 2,000 to 3,000 plate numbers.
Patrol car computers access a database of stolen vehicles, and the system checks every license plate it captures in real time, Cantrell said. If a license plate matches data in the database, the system notifies the officer.
Agencies can create a hot list of license plates of interest, such as vehicles operated by people of suspected terrorist activity or felony violations, Combs said, adding that “it lets us run many more license plates than an officer encounters manually in a shift to get violent criminals off the street.”