Hendon Publishing - Article Archive Details
Automatic license plate recognition
Written by Thomas M. Manson
By Thomas Manson
We first covered automatic license plate recognition (LPR) systems in the December 2006 issue of LAW and ORDER. LPR was heralded as an innovative technology that could change the way law enforcement did business by providing officers instant information about their contacts via the automobile license plate. By that time, only a few agencies had ventured into the LPR realm. Many of these innovators were secretively deploying the technology with specialized anti-theft or narcotic units. A few agencies were more vocal about their adoption of the technology, inviting the media to witness their new crime-fighting capabilities.
In 2006, many companies claimed to have an LPR solution, one that would free an officer from “running a plate” by having to type in the tag or plate number by hand. Vendors claimed that thousands of plates could be “read” during a single shift. Agencies were told that catching stolen vehicles would be easier because so many plates could be processed; racial profiling complaints would also decrease because the systems were focused on the plate and not the driver.
At the time, few agencies had even heard of the technology, fewer still had ever seen it, and fewer than 50 agencies had made actual purchases. In this article, LPR topics that were raised in 2006 are reviewed, and new developments in LPR and law enforcement’s application of the technology are detailed.
Even though LPR technology is widely available from several vendors, the process for most U.S. officers to check a license plate or tag number is the same as it was 10 years ago. If an officer sees a vehicle or plate of interest, he “runs” the plate by either radioing in the number to dispatch or by typing the number into a mobile computer. By either method, the number is checked against a centralized database to see if the plate or car is stolen. The information is then relayed back to the officer, who proceeds with a traffic stop or continues on to another vehicle. An experienced officer might reasonably be expected to process 150 license plates over the course of a shift. But with an LPR-equipped vehicle, the process is much, much faster.
While today’s LPR products vary by manufacturer, the general process and the equipment is very similar. In several systems, the cameras and other components parts are identical, with the real differences being in the quality of the optical character recognition (OCR) software, the user interface, and the back-end management of data.
The systems are usually composed of one or more infrared and color video cameras (early LPR models featured either one or the other, but now both are common). They are typically mounted on the top or sides of patrol vehicles in nondescript black boxes. Each frame from the camera’s video is processed by a computer that “looks” for a license plate-sized object in the frame.
If a plate is identified, it is analyzed by OCR software to determine its sequence of numbers and letters. This alphanumeric string is then compared against a database of “wanted” numbers. This database typically contains wanted vehicles or plates, any number of other databases of wanted felons, outstanding warrants, or scofflaw violators. Manually entered numbers for AMBER alerts can also be included.
If the plate isn’t captured or if no “hit” is registered, the system disregards the information and proceeds to look for another plate. If a hit is registered, the officer is notified with an audible and/or visual alert. Most LPR systems display the image of the hit and return additional information from the onboard database. This data typically include plate number, address, description, reporting agency, and violation. The entire LPR process is totally automated, requires no officer intervention and takes less than a second to perform.
The Difficulty in
In 2006, 50 vendor companies responded to a major agency’s requested need for LPR systems. With OCR software available off the shelf (once found only in super computers, the software is today bundled with $50 flatbed scanners), and infrared cameras available at RadioShack, it seemed like an easy task to put the two together, mount the system on a patrol vehicle and compare the gathered data against a database.
Other obvious obstacles to overcome—wireless data transfer and mobile computing—had all been successfully dealt with by most agencies, so on paper, the solution didn’t appear to be that difficult. And with the major players charging $20,000 for a single LPR deployment, an innovative company capable of bringing the cost down to say, $10,000, would have a clear path for massive growth. But it wasn’t that simple.
If a similar sized agency sent out a request for proposals for LPR systems today, the list of vendors willing to complete the paperwork would be considerably fewer than 2 years ago. Why? Because most companies can’t do it. Cobbling together a piece of technology here and another piece there might look great on paper, but it rarely stands up to the rigors of law enforcement work.
Many agencies during testing and trials were willing to give the “little guy” a chance at the contract, but too often they were eliminated by not being able to capture the same number of plates as the more established companies. Or they were unable to accurately process the numbers and letters of the plates they did capture.
With accuracy rates surpassing 90% for collected plates, and successful duty shifts showing thousands of plate captures, the cards were effectively stacked against companies not equipped with a history of successful image processing or other related experience. Two years ago, buying an LPR system meant taking a leap of faith that the company and the technology would continue to work in the near future. Today those fears are put to rest; only the strong (and effective) have survived.
While 1.2 million cars are stolen in the U.S. every year, only one in eight cases are cleared by arrest. As such, the first noted applications for mobile LPR was for finding stolen autos. Many larger police departments in the U.S. have teams dedicated to the recovery of stolen vehicles, and with specialized training, officers become adept at recovering them. Keeping an eye ever peeled for popped out side windows, pry marks and other telltale signs of theft, a seasoned veteran can rack up a sizable list of recovered autos in a few years.
But an LPR system can pull stolen autos off the road one after another. In one early trial in a major northwest city, the trial had to be temporarily suspended because the LPR system was overloading the department’s ability to arrest drivers and haul away the stolen vehicles.
Not every jurisdiction has a large vehicle theft problem. Stolen vehicles may be a relatively small area of concern for many agencies. Recently, new adaptations for LPR have come to the forefront.
Everyone needs a place to park. If you are downtown, even in Smalltown, U.S., you need a place to park. But if you leave your car in one place too long, you’ll likely get a parking ticket. And while non-moving infractions don’t usually receive much attention from law enforcement administrators, they have begun to catch the attention of city and municipal leaders. Why? Because while rather small individually, the aggregate of all of the unpaid fees adds up. Several agencies have invested large sums of money in complete LPR systems only to see them pay for themselves in recovered fines and seized monies in only a few days. One agency was able to pay for its system with only a few hours of recovery work.
Innovative uses for LPR are being thought up on a regular basis. One of the more interesting deployments is on school buses, which carry millions of children to school and back home every day. Our most valuable assets are often placed in great peril by careless or worse yet, reckless drivers, who pass from both head-on and from behind while the STOP arm of the bus is extended.
In one deployment in Syracuse, NY, an LPR system by Elsag North America captured nearly 100 illegal passes of a single bus in a only a few weeks. Traditionally, these violators would have continued on without any potential for citations or arrest unless a patrol vehicle just happened to be nearby. Now these violators are sent a summons to appear or pay a stiff penalty, a photo of the violation is included with the summons.
Another unique application for LPR is its adaptation on the back of a street sweeper. In Chicago, a Canadian-based, Genetec LPR camera system has been added to a street sweeper, which captures the plate number and a picture of the parking violators who have failed to move their vehicles off the roadway on posted days for street sweeping. This seemingly inglorious LPR application has the ability to not only increase revenue from parking infractions and increase compliance during city cleaning schedules, it also has the ability to tie its database to local law enforcement, allowing officers to find vehicles that have been ditched on the street rather than in parking lots or abandoned areas.
The Value is in the Data
LPR systems, regardless of the manufacturer, share several similarities; they use infrared video cameras and compare the captured data against a known database of wanted numbers. LPR systems also have the ability to populate databases with their collected data. In early 2006, few law enforcement agencies were interested in what they could do with the large amounts of data collected by LPR systems. That an officer could process a thousand of plates a shift and find a few stolen vehicles in the dark was more than enough to justify the systems. At the end of the shift, many agencies just simply deleted the non-stolen reads.
The rationale for this was three-fold. First, the data wasn’t viewed as important; if there had been a hit on a stolen vehicle, it would have been dealt with the moment it was recognized, so there was no need to keep the non-stolen data.
Second, the non-stolen data was viewed as important but potentially volatile and therefore should be deleted as soon as possible. Individual advocacy groups like the ACLU and civil libertarians had at this point given LPR systems their weakest, tacit approval. The systems were “equal opportunity employers,” looking only at license plates and not at the race of the person behind the wheel. But as with any collection of data about individuals, there was the potential for abuse. It was probably just as well to delete the data. A third rationale was simply that the agencies didn’t know what they could do with the non-stolen data, so why save it?
Today, the applications of LPR for the officer on the street have expanded beyond simply looking for stolen vehicles or plates. Databases for scofflaw violators, wanted persons, narcotics interdiction, and gang and terrorist watchlists are routinely being integrated into LPR systems. The trend to connect various databases to LPR systems has inspired many officials to keep the data collected by these systems for future data mining.
Because a majority of criminal behavior is connected to the use of automobiles, the ability to identify the time and location of the license capture becomes a valuable tool for many aspects of law enforcement. If, for example, a burglary is called into
9-1-1 and the units arriving on scene and those in the surrounding area are equipped with LPR systems, their collective data could create both a suspect list and a potential witness list for the crime.
Similarly, if a series of burglaries occurred, the data collected from LPR systems in the area at the time of the burglaries could be combined, or “mined” to see if any common vehicles were in proximity to the crimes and create a narrower suspect list or better profile. It is not difficult to foresee a future where the data collected from LPR systems becomes more valuable to agencies after the fact than to the officer’s who originally collected it.
Fixing the Mobile Camera
The mobile LPR systems that are the focus of so much interest currently in the U.S. are actually derived from fixed systems that have been in use Europe and other parts of the world for decades. Fixed LPR systems were first deployed in the U.S. along the borders and in parking lots. But interestingly, as mobile systems are being improved and increasingly implemented, more agencies are returning to LPR’s roots and deploying fixed systems.
Technologically, deploying a fixed LPR system is much easier than a mobile system. On the U.S. border, at a port of entry for example, the lanes of travel are well defined and can be augmented with lighting systems, which generally guarantees the optimal conditions for capturing a license plate. Mobile LPR, on the other hand with its unpredictable conditions of road, weather, lighting, angle of attack and vehicle speed variables is a much greater challenge. Two years ago, mobile LPR with its high-speed captures was being touted by manufacturers and agencies alike as the technology of choice, but today, virtually every manufacturer is highlighting its ability to “fix” its system to poles and traffic signals.
When it comes to shear volume of license captures, fixed LPR systems win hands down. Mounted on a traffic signal at a busy intersection, a well-place, fixed LPR system can capture nearly 100% of the traffic within its field of vision. Even the best mobile systems can’t claim that distinction. But there’s one drawback: there’s no patrol vehicle set to respond if a vehicle of interest passes through the monitored intersection.
The advantage of the mobile LPR system is one of immediacy. If an officer is notified of a hit, he sees a color image of the vehicle and proceeds with a traffic stop. The reaction is immediate. The ability to react to the data more effectively gives mobile LPR an advantage that their fixed brethren cannot claim. But when it comes to data mining, both systems are equally important.
If an agency wanted to generate a list of the vehicles that passed southbound away from a major crime scene during a given time window, a fixed system would be best equipped to provide the data, but only a mobile system could have driven around the crime scene itself and processed all of the plates parked and moving about within the area.
Similarly, if a “geo-fence” has been established around a school, preventing sexual predators from driving within a set distance of the school, a fixed system would be able to provide immediate alerts to agency official of an incursion, but only an officer in a mobile LPR equipped vehicle would be able to effect an arrest of the subject at the scene.
As LPR technology is more widely adopted, law enforcement agencies and departments will likely experiment with both fixed and mobile systems to see which makes the most sense for their community and criminal activity.
One of the more interesting trends in the LPR industry is the buying and selling among several of the leading companies. No fewer than four of the largest LPR vendors have experienced some sort restructuring in the past 18 months. This is indicative of the relative youth of the industry and the intense interest the field is generating among investor groups.
Elsag North America, maker of the prevalent MPH-900, is now wholly owned by Italian-based Elsag, after its former partner Remington was purchased by a New York investment firm. Tennessee-based, PIPS Technology was purchased in August 2007 by well-known vehicle supplier and Illinois-based Federal Signal in its interest in diversifying into different law enforcement sectors.
Platescan, formerly a division of Civica Software was separated to form its own company after interest and likely funding from the CIA. And Canadian-based AutoVu was acquired by another Canadian firm, mega-company Genetec. All of the purchases, positioning and restructuring reveal a growth industry with the perceived ability to generate significant revenue.
At a time when most of the LPR companies are pursuing hard-mounted, semi-portable products, one company is going in a different direction. In 2006 INEX Technologies of Knoxville, Tennessee, and Zamir Recognition Systems of Jerusalem, Israel, joined to form INEX/ZAMIR.
Known internationally for their fixed high-speed LPR applications (tollways and parking lot access) INEX/ZAMIR has opted for a highly portable, handheld LPR design. Instead of requiring an expensive vehicle upfit, one cable of the model LY-300 simply plugs into a DC power port and the other cable into a laptop. Beyond portability, another added benefit is cost; the LY-300 is nearly half the price of the other LPR systems.
In the 2006 article, LPR systems were predicted to be integrated with other existing patrol systems. This has already come to pass with the integrations of in-car video with LPR systems. The Platescan LPR system now integrates with Panasonic’s Toughbook Arbitrator mobile video recording system. When not in use to record video of traffic stops, the system can be set to search for license plates.
Other integrations will continue, as well as new adaptation for the systems themselves. And as the industry settles and more agencies make purchases, the cost of the technology will continue to fall. Today LPR systems are currently deployed in less than 3% of law enforcement patrol vehicles in the United States, but through their own merits, they are quickly changing the way law enforcement does business. n
Thomas M. Manson is the owner of Police Technical LLC and the technology editor for LAW and ORDER magazine. He speaks nationally on technology and law enforcement. He can be reached at firstname.lastname@example.org.
Published in Public Safety IT, Sep/Oct 2008
Rating : 10.0
Related ProductsALPR (Automated License Plate Recognition)
Click to enlarge images.