Automatic
License Plate Recognition
By Dennis J. Lau
What is Automatic License Plate Recognition?
In an era of increasing threats to public safety and shrinking city budgets,
judicious application of modern technologies are a force multiplier for police
agencies who need to do more with less. Automatic License Plate Recognition (ALPR)
is a case in point. ALPR uses cameras to capture digital images of license
plates, then a computer to convert plate image into alphanumeric characters.
That information, also referred to as “plate code”, can then be stored in a
database and compared to other databases.
Automatic License Plate Recognition got its start in the 1990s when the
British responded to a series of bomb attacks by members of the Irish
Republican Army by developing London’s “Ring of Steel.” Using fixed mounted
cameras in strategic locations, authorities capture the plate code of every
vehicle entering London. (Coaffee, pp. 204-205) ALPR is used in parking
revenue systems by recording cars as they enter and exit the lot, in
Intelligent Transportation Systems to conduct Origin-Destination and Travel
Time Studies by calculating the time it takes for a vehicle to pass between
two points, in Cordon Studies to determine traffic patterns in and out of an
area, and automatic vehicle tolling systems. (Rossetti, 2001)
State and local law enforcement has begun deploying vehicle based systems.
With current computer processing speeds, police cars can be equipped with up
to four cameras running simultaneously, checking the results from each plate
it “sees” against a criminal data base, and alerting the officer with a
preliminary match or “hit” within a couple of seconds. These systems can
accurately read plates at speeds up to 160mph in closing and greater than
75mph in passing. (Senn, 2008)
How does it work?
Automatic License Plate Recognition can trace its beginnings to character
reading machines used in mail sorting over 50 years ago. An ALPR system has to
locate the license plate in an image of the vehicle and then perform optical
character recognition (OCR) on that plate. It is comprised of a camera or
cameras, an illumination source - often infrared (IR), image processing
software, a host computer, a power supply, and network connectivity should the
system need to communicate with external databases for its operation.
The challenge from converting mail reading technology, to grabbing characters
off of fast moving vehicles, at random and variable angles and light
conditions, is achieved by clever mathematics and modern computer processing
power. Fuzzy logic is often used to look for where the license plate should be
based on assumptions of where a plate should be located, and edges, colors and
textures that would correspond to images of license plates. Algorithms, neural
networks and fuzzy logic, segment out the characters in the image and covert
them to alphanumeric plate code. (Jia, He, & Piccardi, 2004)
In a law enforcement application, plate code is commonly compared to a
database of wanted persons and stolen vehicles. If there is a positive match,
the system alerts the officer for further action.
How is it used in law enforcement?
Without a vehicle based ALPR system, a police officer check plates by calling
them out over the police radio for a dispatcher to check and report back to
the officer. Officers who have Mobil Data Computers (MDCs) installed in their
vehicles, that are wirelessly networked to their department’s Computer Aided
Dispatch (CAD) system, can play license plate bingo directly on their MDC
without dispatcher assistance. A technically competent officer can “run” about
150 plates in a normal shift. With an ALPR system, the same officer can expect
to have checked 3,000 -5,000 plates in a single shift. (Plourd, 2006)
The technology allows the same work to be completed with fewer officers.
Retired New York Police Lieutenant, Pat Fitzgerald had this to say: “It's very
efficient… for example, the city of White Plains hired one officer for 8 hours
on over time to run plates (without a reader) in a particular parking lot. I
did this same lot in 20 minutes.” (Senn, 2008) Other agencies have had great
success assigning them to Auto Theft Task Forces. The Pennsylvania Auto Theft
Prevention Authority has seen a dramatic drop in auto theft. When it began in
1996, 53,000 auto thefts were reported that year. When the authority began
using mobile ALPR “in earnest” in 2007, the annual auto theft rate dropped to
28,000. As the word gets out, they’re hoping for continued declines. (McKay,
2008)
While the initial driving force behind law enforcement ALPR was to locate
stolen cars, the enormous data collection activities of these systems offer
public safety benefits far beyond the efficient location of wanted persons and
stolen vehicles. When fixed and mobile ALPR systems are connected to robust
“back office” hardware and software, ALPR data, which includes the plate code,
date, time, location and color picture of every single car it saw, can be
stored in a single searchable database. An industry consortium of ALPR system
vendors are adopting data format standards so ALPR data from the different
vendor’s systems can be shared and searched. (Reuters, 2008)
Regional ALPR data can be “mined” to solve crimes by placing the same car at
multiple crime scenes. Driving an ALPR car around the area of a crime can also
aid in placing a suspect at the scene and locating witnesses. LASD recently
solved a serial rapist and a murder case by using ALPR data to locate the
suspects. (PRNewswire, 2008) Fixed ALPR can be used as one layer in a
multi-layered security approach, to protect critical infrastructure. The same
plate seen at multiple sensitive key infrastructure locations could generate
leads for investigation by anti-terrorism task forces, which causes some
heartburn for the ACLU.
What are the civil rights issues?You can bet that the ACLU has not
been silent on this issue. Different ACLU offices have freely offered their
wisdom. "The NYPD has proposed the blanket, indiscriminate videotaping of
millions of people," said New York Civil Liberties Union Associate Legal
Director Christopher Dunn. "The NYPD should not be spending $100 million of
public money to track law-abiding New Yorkers." (Gendar, 2008)
While not public safety experts, they freely comment on perceived governmental
intrusions. Jeff Gamso, legal director of the American Civil Liberties Union
of Ohio said, "The scanner’s gaze is too wide and it’s an infringement against
the innocent drivers whose plates get captured… Using (ALPR) to scan all
license plates is a civil rights violation and could lead to government abuse
of the information… I think they should just knock it off." (Associated Press
, 2007)
The Washington State Patrol started checking all the plates of vehicles
boarding ferries. They check them for AMBER alerts , reported stolen vehicles,
felony wanted persons, and suspected terrorists. ACLU Seattle chapter
spokesman Doug Honig said, “The (ACLU) has no problem with the camera system
being used for flagging stolen vehicles, potential child abductors and wanted
felons” but not terrorists. (Port Orchard Independent - Opinion, 2008)
The running thread is concern is how long the information will be stored, how
it will be used, and who will have access. Ultimately the courts may have to
weigh in on these plain view observations and the need for public safety vs. a
reasonable expectation of privacy.
How does the equipment from different vendors compare?
Law enforcement agencies with limited technical expertise are often challenged
to select technology based equipment, systems and services. Trade shows and
vendor claims can be problematic to sort through. Recognizing this dilemma the
U.S. Department of Homeland Security established the System Assessment and
Validation for Emergency Responders (SAVER) Program to provide local agencies,
objective tests of comparable equipment. SAVER focuses on two questions: “What
equipment is available?” and “How does it perform?”
In 2008 SAVER issued a series of reports on mobile license plate recognition
systems. Its focus group recommendations set evaluation criteria, comparison
methodology, and selected four vendors for evaluation. Systems were evaluated
on a 5 point scale in the areas of capability, usability, deployability and
maintainability. The PAGIS system by PIPS Technology scored the highest with
an overall score of 4.4, followed by MPH-900 by ELSAG North America with an
overall score of 4.0, and PlateScan by Civica Software with an overall score
of 3.5. CarDetector by Vigilant Video received the overall lowest score of
3.1. (Engstrom, 2008)
If a local agency is deploying new ALPR technology, the consideration for
regional compatibility with ALPR systems used by other local agencies should
be considered. A good first step would be to talk to their other local
agencies that are already using ALPR systems and learn from their experience.
ELSAG is well established in the east, while PIPS has a strong following out
west. I believe that ALPR is already an effective tool for law enforcement,
and like all technology, it continues to improve rapidly with time.
ABOUT THE AUTHOR
Dennis J. Lau is a law enforcement official in Southern California. He
completed this paper in conjunction with his course work while pursuing his BA
in Criminal Justice Management.
References
Associated Press . (2007, July 30). ACLU objects to plate scanners in police
cruisers. Retrieved March 22, 2009, from The Chronicle-Telegram: http://www.chroniclet.com/2007/07/30/aclu-objects-to-plate-scanners-in-police-cruisers_122/
Coaffee, J. (2004). Rings of Steel, Rings of Concrete and Rings of Confidence:
Designing out Terrorism in Central London pre and post September 11th.
International Journal of Urban and Regional Research , 28 (1), 201-211.
Engstrom, C. (2008). Mobile License Plate Recognition Systems Assessment
Report. Washington D.C.: U.S. Department of Homeland Security.
Gendar, A. (2008, August 12). ACLU sez city plan to scan license plates bum
steer. Retrieved March 22, 2009, from NY Daily News: http://www.nydailynews.com/ny_local/2008/08/12/2008-08-12_aclu_sez_city_plan_to_scan_license_plate.html
Jia, W., He, X., & Piccardi, M. (2004). Automatic License Plate Recognition: A
review. Proceedings of the International Conference on Imaging Science,
Systems and Technology (pp. 43-48). Las Vegas: CSREA Press.
McKay, J. (2008, April 9). License Plate Recognition Systems Extend the Reach
of Patrol Officers. Retrieved March 22, 2009, from Government Technologies:
http://www.govtech.com/dc/articles/282014
Plourd, K. (2006, April 18). Camera system aids Long Beach police. Retrieved
March 21, 2009, from CSULB Online 49er: http://www.csulb.edu/~d49er/archives/2006/spring/news/v12n104-camera.shtml
Port Orchard Independent - Opinion. (2008, July 30). WSP plan irks terrorists
and their enablers. Retrieved March 22, 2009, from Port Orchard Independent:
http://www.pnwlocalnews.com/kitsap/poi/opinion/26023824.html
PRNewswire. (2008, November 10). Los Angeles County Advances Public Safety
with Federal Signal Technology. Retrieved March 22, 2009, from Officer.com:
http://www.officer.com/web/online/Off-The-Wire/Los-Angeles-County-Advances-Public-Safety-with-Federal-Signal-Technology/27$44070
Reuters. (2008, January 15). PIPS Technology Forms License Plate Recognition
Industry Consortium. Retrieved March 22, 2009, from Reuters: http://www.reuters.com/article/pressRelease/idUS220179+15-Jan-2008+PRN20080115
Rossetti, M. D. (2001). Applications And Evaluation Of Automated License Plate
Reading Systems. Proceedings of the Intelligent Transportation Society of
America 11th Annual Meeting and Exposition. Miami.
Senn, P. M. (2008, July 8). Picking Out Plates. Retrieved March 21, 2009, from
Officer.com: http://www.officer.com/article/article.jsp?id=30987&siteSection=20
