ALPR: Taking License Plate Recognition to a New Level
Written by Dale Garrison   

For several years, Automated License Plate Recognition (ALPR) technology has expanded identification capabilities for everything from law enforcement to government use and private business. With dramatic increases through computing power, wireless technology, and analytics, the ability to identify vehicles and cross reference that information with other data is expanding at an impressive rate.

ALPR is not new. As early as the 1990s, observers realized that computerized optical character recognition (OCR) could be used to identify license plates. Police departments were especially eager to exploit the concept and were among the early adopters of the technology. In Europe, large systems were even developed that attempted to create a virtual wall around cities like London.
Early ALPR was similar to camcorders or closed-circuit television cameras, but it has evolved tremendously. One development includes cameras with infrared illuminators. Unlike closed-caption display sensors, infrared illuminators function equally well day or night and regardless of weather conditions.
Seth Stores, global business manager for 3M, said another turning point in ALPR technology has been the advent of wireless Internet. “This allowed mobile ALPR cameras to send data to backend software and servers in near-real-time,” he said. “Adding online functions to support the technology increased the benefits that ALPR can provide.”
Randal Raszick sees some of the biggest developments in the evolution of ALPR software. A sales engineer for PlateSmart, he describes the company’s efforts to perfect comprehensive image analysis, a more sophisticated cousin to OCR. “We developed our technology not just with optical character recognition, but in image analysis,” he said. “We get a lot more information out of a vehicle passing than other systems. It’s also very flexible and adaptable.”
PlateSmart’s system is purely software and works with almost any hardware, including older analog cameras. “It’s an open architecture that does true image analysis using off-the-shelf cameras,” he said. “It looks everywhere in the frame—there’s no ‘sweet’ spot where the license plate has to be to be read.”
Stores at 3M also noted the growing sophistication of ALPR software and analytics. He cited 3M’s new ALPR software, 3M Plate Alert Analytical ALPR. “It goes beyond ALPR software by using data fusion analytics to uncover relevant connections that give users actionable insights,” Stores said. “Plate Alert software can compile ALPR data with data from various sources, such as 911 calls, pawn data, police reports, motor vehicle data, and more. It can then almost instantly find connections between data points to deliver solid, meaningful leads.”
Plate Alert also incorporates what 3M calls Pattern Management Alerts. These alerts automatically push notifications of activity to users, helping them take action faster. “Notifications are sent when the system detects activity that matches a pre-defined pattern, and alerts proactively provide relevant information so almost all the work is done before someone even looks at the data,” Stores said.
Other advances continue. Raszick at PlateSmart said that company is preparing to launch a high-definition system that will allow recognition of license plates from two lanes. He sees other announcements likely this year. “We’re seeing advances about as fast as we can perfect them,” he said. “It’s very exciting.”
Serious Start
Part of the technology’s evolution is evident in its history. One story involves John Chigos, CEO of PlateSmart, who was living and working in New York in 2001. He had an appointment in the Twin Towers the morning of Sept. 11, 2001. He had to miss the meeting, but several of his friends and associates died during the terrorist attack. Later, when he learned how some of the terrorists had been identified but had driven to flying lessons and other locations without being detected, he began searching for technology to help keep this from happening again.
“One of the things he realized is that, especially in America, license plates are already available with a large database you can tie to,” Raszick explained. “And 75 to 80 percent of our crime is committed with a vehicle. He realized that license plate technology is the best resource for this.”
One big challenge in this country involves the lack of standardized plate design, making OCR difficult. In Europe, where ALPR has been more widely used, plates tend to be more standardized with relatively simple black and white designs—no swirling art behind or around plate numbers. Realizing that the systems developed and used in Europe would work poorly and be expensive in the United States, Chigos sought another option. “The whole point for this is to have widespread access and deployment,” Raszick recalled. “He had to look for something else.”
Older systems were also limited in other ways. Development often followed one of two tracks: expensive mobile systems that could be used in a patrol car, or huge “ring of steel” networks such as the one in London, England where the city is enclosed by a virtual wall of expensive video cameras.
“Only a large city can afford that,” Raszick said. “Even a city of one million people can’t afford that.”
Full-Time Options
Mobile systems avoid some of those issues but only operate when the vehicle is manned and in use in a suitable location. “When the officer is out of the car or doing something else, it’s not getting used,” Raszick said. “That $25,000 system is not doing any good.”
That’s why Raszick sees a major trend in ALPR toward cameras on poles at “chokepoints,” gateways where a fixed camera operates 24/7. Linked to a base via Wi-Fi, the software can be installed on a laptop and run reports nearby, working with virtually any camera that meets minimum specifications, accessing almost any visual recording device on a network. “It’s fully networked with open architecture,” he said. “With a system like this, I can pull in video from anywhere in the world and check it in real time.”
That has led some to shift away from mobile systems to strategically placed camera locations. But rather than attempting to cover every foot of a perimeter, the concept focuses on key points where vehicles pass, saving immensely on hardware costs.
In many cases, business users are not screening licenses by themselves, but instead are sending data to local police. “What I see more customers doing is simply sending plate numbers to police for lookup,” Raszick said, citing hospitals and similar locations. “The hospital won’t know if that’s someone on an Amber Alert or wanted for assault, but the police will. This provides people in a homeowners association, elementary schools, wherever, to have the capability of having law enforcement there even through there’s no one actually there. That’s one of the big trends I’ve seen.”
And like modern cars, which are increasingly improved with software updates, such systems can be enhanced without buying new cameras or other hardware. “As technology improves, we keep up with it,” Raszick explained. “We’re not locking them in to a $5,000 camera that they have to upgrade if the software upgrades. That doesn’t happen. If they have the infrastructure in place, all they do is add our software. They can keep what they have and still upgrade.”
All of this explains why some of the most widespread uses are now outside of law enforcement. ALPR is a perfect fit when private businesses want to monitor cars in a worksite or another location, even for something as simple as ensuring legal parking. Using a database of employee auto license plates, such monitoring is relatively simple and avoids issues of privacy because those being monitored are already listed as employees in a limited database, not part of public monitoring. The systems are fast and accurate and can be used for everything from small factories to multi-campus hospitals, police agencies, or government and military locations.
Another advantage of these systems is that they are truly automated, requiring minimal training before use. “The nice thing is, it’s easy to set up and usually takes less than 30 minutes to become operational,” Raszick said.

Automated license plate recognition technology is a good fit for private-business applications, such as monitoring vehicles at a worksite or parking garage. Photo courtesy 3M.
Other Advances
The continued evolution of the technology is still under way. In May of this year, PlateSmart introduced software that can determine the make and model of a car. By the end of this summer, they’ll be able to analyze images in order to determine the type and color of vehicle, license plate number, and state. “You’ll know whether it’s a black BMW with a New Jersey license plate, a car, pickup, or motorcycle,” Raszick said.
Again, the key to this capability is that the software is analyzing almost everything in the image, not just license plate numbers. Systems can even be trained to “look” for behavior patterns such as a vehicle circling a block where a bank is located, or parking too long at an airport terminal curb.
“The software, technology and object analysis leads into analytics,” Raszick added. “The more points of information you give to your user, whether it’s a bank or highway weigh station, the more it lets you do something called multiple-factor verification.”
That level of analysis is a huge leap for the technology, expanding where and how it can be used as well as its sheer capability. A simple example involves an auto inspection station where individuals may seek to get a permit for a car that they know won’t pass an inspection. A trick used by some auto owners is to temporarily swap plates with a car that can pass. With ALPR, the inspection station can simply run the plates to see if they are on the auto they should be on. The trick won’t work, even if an employee at the station tries to interfere.
“What if a Honda with a given license comes in for an inspection, but the license is registered to Hyundai?” Raszick asked. “With an ALPR, we’ll catch them. Even the technician can’t collude. The system checks on the registration record and sends an alert.”
The same analytics can be used on toll roads to replace pass systems or for similar applications. “There’s a tremendous array of ways our system can be used,” Raszick added. “The neat thing is, we have customers coming to us asking, ‘What else can we do?’ We’re getting a lot of ideas that way.”
One case involved applications at a large ocean port where cameras had been installed to count vehicles. The customer soon returned and asked if the system could also count tractor-trailers and, of course, the answer was Yes. But that’s only the start.
Even the relatively simple capability of networking changes the game considerably. Basically, the systems are connected by a computer network and can be shared over the Internet, making them available virtually everywhere. Raszick recalled a demonstration to Department of Navy investigators who had been seeking for years to connect cameras at an intersection. “We stopped the presentation and in about eight minutes we had it set up,” he recalled. “They had given up on the technology until they saw this.”
Working with all that data can often be intimidating, difficult, and time-consuming. Here again, software systems offer solutions. For example, 3M’s Plate Alert software provides robust tools to allow agencies to efficiently manage their data retention requirements, with strong auditing capabilities to help data usage compliance and multiple controls to limit who can see what data.
“The software can fuse together diverse data sources—sources such as ALPR data, 911 call data, pawn data, incident and arrest records, and more—and integrate them into a single, unified view to create a more complete, informative, accurate, and useful picture,” Stores said. “And by linking and simultaneously evaluating numerous common characteristics across multiple data sources, the software can uncover unexpected relationships.”

Today’s ALPR systems can utilize image analysis in addition to optical character recognition. They are also easy to use, requiring a minimal amount of time for training and setup. Photo courtesy PlateSmart.
Outside Issues
Similar advances of forensic technology often face major questions from privacy advocates and others. Although there are some questions regarding ALPR, they have been minimal compared with areas such as facial recognition or even DNA. For one thing, license plates are already public and the technology is simply facilitating the ability to look up connected, public information more quickly.
“All we’re doing is enabling the look up,” Raszick noted. “Is that the vehicle from an Amber Alert? Is that car coming into the lot a customer? Is that someone on a compulsive gambler list at a casino? The potential is tremendous.”

About the Author
Dale Garrison is a freelance writer in Liberty, Mo.


< Prev   Next >

Image Clarification Workflow

A FEW WEEKS AGO, I received a call from Ocean Systems asking if I would like to beta test their newest software—ClearID v2.0 Image Clarification Workflow. The new progam has filters that were designed for use with Adobe’s Photoshop graphics-editing program.