By Captain Russell A. Neville, Information Technology Management Section Commander, Cincinnati, Ohio,
utomated License Plate Recognition (ALPR) technology is one of the hottest technologies in the law enforcement community today. With the increased deployment of this technology around the United States and worldwide, the Cincinnati, Ohio, Police Department (CPD) has shown how the technology can be deployed and shared effectively across multiple law enforcement agencies to increase overall public safety in the greater metropolitan area of Cincinnati. Although much of the media attention revolves around the recovery of stolen vehicles, the CPD has demonstrated that the technology provides many more benefits to public safety practitioners and the residents of the communities they serve. Automatically checking over three million vehicle license plates in its early stages, the system has already generated incredible results that extend the capabilities and the effectiveness of the CPD’s efforts to improve public safety.
The CPD is now leading the development and deployment of a large regional ALPR network covering 12 counties in the tristate Southwest Ohio/Southeast Indiana/Northern Kentucky (SOSINK) region. All ALPR data in the SOSINK project will feed into a central database that is accessible to more than 200 law enforcement agencies across the region for what will be an unparalleled model for the broad regional deployment of ALPR technology.
This article provides a brief explanation of the technology, an overview of the system deployed at the CPD, the benefits realized to date, future plans for expansion, lessons learned, and considerations for other agencies looking to deploy ALPR technology in their jurisdictions.
ALPR systems are designed to automate the process of checking license plates, a duty that officers already perform manually on a regular basis. Highly specialized cameras, either mounted on a patrol car or fixed along a roadway, use infrared illumination to make license plates visible at any time of day or under any weather condition. The cameras have two lenses to capture two images: a color image of the vehicle to create an evidential record and an infrared image of the license plate for identification. Fixed ALPR systems are more strategic as they provide continuous monitoring of high-traffic areas, whereas mobile systems are more tactical, allowing for rapid deployment into areas of high rates of known criminal activity and also for the routine collection of data for investigations.
The infrared image of the license plate does not “see” the graphics and color schemes used on the plate; instead, it produces a clear black-and-white image that the system can then interpret accurately using optical character recognition (OCR). The plate interpretation is then checked against various databases to determine if the vehicle might be of further interest. The CPD currently has set up the system to look for vehicles whose data have been entered into the National Crime Information Center (NCIC) database as well as be-on-the-lookout (BOLO) notices, local wants and warrants, Amber Alerts, and more. In the event a mobile ALPR system receives a database match, the officer using the system is provided an audible and visual alert with all pertinent data; in the case of a fixed camera, a notification may be sent to dispatch for appropriate action. All of the data collected are then aggregated into the Back Office System Software (BOSS) for investigative use, data analysis, mapping, and sharing with other agencies.
Cincinnati’s first mobile ALPR system was installed in the fall of 2007, with the primary goal of the project being nothing more than to evaluate this relatively new technology. The CPD met with several vendors and chose to field-evaluate Federal Signal’s PIPS Technology three-camera P362 mobile unit. The project quickly gained traction, getting attention throughout the department, the media, and the community. Within the first five months, this one system had read more than 300,000 plates and found over 8,000 vehicles of interest to law enforcement agencies, resulting in the arrest of about 300 suspects. Additionally, BOSS’s investigative potential began to show, being credited with helping to apprehend a homicide suspect, a bank robber, and a copper thief.
Based on the results of the first five months, it was quickly determined that the project should be expanded. An additional eight mobile systems have since been deployed, and the success has continued to grow. Surrounding agencies began seeing value in the system and began purchasing systems of their own. It was at this point that the CPD decided to take on a project within a project. Using data-sharing capabilities within BOSS, the CPD began hosting data and sharing these data with these other agencies, including the Hamilton County Sheriff’s Office and the Green Township Police Department. The additional five mobile systems deployed at these agencies began to feed data into this hosted and shared BOSS, and the CPD immediately began to realize the benefits of a more comprehensive regional approach to intelligence gathering.
The database has now logged more than three million license plates and is used daily for investigative purposes. Many local and regional law enforcement agencies have been given access to this database, including the Terrorism Early Warning Group (TEWG), the U.S. Federal Bureau of Investigation (FBI), and the U.S. Drug Enforcement Administration (DEA). In January 2009, the CPD became the first agency in the United States to deploy the new Slate low-profile mobile system on one of its Vortex Unit cruisers (see figure 1). This system has shown increased performance and a significant size reduction of the cameras, all with an overall reduction in cost. Twenty-seven additional Slate mobile units are in the process of being deployed throughout the Ohio counties in the SOSINK region thanks to Law Enforcement Terrorism Prevention Program investments.
|Figure 1. A low-profile mobile ALPR system mounted on one of the CPD’s Vortex Unit cruisers.|
Photo by Heather Whitton, ALPR project manager, CPD
With the increased deployment of ALPR technology around the United States and worldwide, the Cincinnati Police Department has shown how the technology can be deployed and shared effectively across multiple law enforcement agencies to increase overall public safety in the greater metropolitan area of Cincinnati.
Realizing the System’s Potential
When the CPD began its ALPR project, the potential investigatory value of this technology was not yet fully understood or appreciated. Today, ALPR technology is seen as a powerful tool to help protect citizens and communities and to acquire beneficial intelligence data. CPD officers and investigators use the BOSS database on a daily basis to investigate Uniform Crime Report Part I and Part II crimes, including homicide and aggravated robbery.
The ability of BOSS to run queries or mine data based on full or partial plate information, time and date, and geographic location is an extremely powerful investigative tool. In many instances, the CPD’s ALPR data have been critical to solving a case, cases that may not have otherwise been solved. Cases closed through the use of ALPR data include a double homicide, an aggravated robbery, a serial burglar, and several vehicle hit-skip auto accidents, to name just a few.
Another powerful feature of the system is the ability to map vehicle data based on the Global Positioning System (GPS) coordinates attached to a vehicle at the moment the system “sees” it. GPS and mapping technologies have helped CPD officers locate numerous suspects and have provided the critical link in solving a double homicide and a gang-related homicide. This information is especially useful for determining the potential whereabouts of suspects when vehicle registration data are nonexistent or inaccurate. Additionally, these data may be used to prove or disprove alibis, look at patterns of vehicle movement for surveillance activities, and look for common vehicles across multiple crime scenes.
The vehicle photographs maintained in BOSS are very powerful and can be used in a variety of ways, such as identifying full license plate numbers based on partial plate information as well as eyewitness descriptions of suspect vehicles, vehicle condition, and other identifiers such as paint color schemes and bumper stickers. The system’s images are also very valuable in lineups and crime bulletins.
Having come to understand the potential investigative power of this system, the CPD now uses its mobile units much more strategically. Mobile units can collect license plate data around a variety of crime scenes or through routine patrol of known trouble spots (high-crime areas, clubs, and so on). They can also be placed ahead of an impaired-driving checkpoint to provide additional information to officers at these locations and increase officer safety by giving them more reaction time. High-priority events such as festivals and parades are also good locations for ALPR monitoring, allowing for some level of screening of vehicles entering these events while also collecting intelligence for later use if needed. Using ALPR systems for this type of intelligence gathering will also help bring the technology in line with several federal grant priorities of homeland security.
Building Out the Vision
With so much success and interest from surrounding agencies, the CPD began to consider a wider application of ALPR technology across the tristate SOSINK region. After holding discussions with other agencies in the region, the CPD and its law enforcement partners concluded that a regional network of fixed ALPR cameras could capture and track regional criminal activity and would likely result in better enforcement, greater deterrence of crime, and improved public safety for the region. This network would cross many jurisdictional boundaries but would be aggregated within BOSS and made available to all 200 agencies in the region. This project would serve as a critical and efficient platform for the prevention of terrorist attacks by supporting and enhancing raw intelligence gathering and regional information sharing among law enforcement agencies. An additional benefit of a coordinated ALPR project with well-defined objectives and multiple agency cooperation would be greater access to increased grant and funding opportunities.
Determining the goals of the program and the type of ALPR system best suited to accomplish these goals is paramount to the program’s success.
The objectives of the SOSINK ALPR project are to produce results in two primary areas that, when combined, create a regional system unlike any other in the United States. The first objective is an increased number of wanted-subject apprehensions. The systems will give local agencies immediate proactive notification regarding vehicles entering the region that are on “hotlists” such as the FBI’s Most Wanted as well as hotlists from the NCIC and other databases. The system will also be able to provide intelligence about locations within the region that a suspect may frequent, thus increasing the chances of location and apprehension. Deployed officers will become exponentially more efficient and effective in accomplishing this goal and will be able to focus their efforts on driving safely and monitoring
The second objective is to collect intelligence and enhance investigations. Collected ALPR data will be available for analyzing traffic patterns of wanted subjects and drug trafficking as well as monitoring subjects on terrorist watch lists and traffic around critical infrastructure. Data analyses will provide increased intelligence for local law enforcement agencies, the DEA, the TEWG, and local FBI field offices. These analyses can also assist in recognizing trends and taking preventive action on early-warning indicators.
The SOSINK project, a collaborative effort monitored and executed with the full support of the SOSINK Region Six Law Enforcement Subcommittee, will involve over 80 fixed ALPR cameras to be located on federal Interstate highways in the region. These cameras will monitor the roadways continually for vehicles of interest while continuing to build out the BOSS intelligence database. The locations of these cameras have been selected based on the available infrastructure and the desire to achieve maximum regional coverage.
To support such a large ALPR network, funding and consideration had to be given to address the need for a centralized and highly secure server, a need that exceeded the capacity and capability of any single SOSINK agency at the time. Several options were considered for handling the large amount of data and for the data mining and analysis that multiple investigators across the SOSINK region would undertake.
After careful evaluation of hardware and software requirements, cost, personnel, electrical, data, and space concerns, the CPD decided to invest in a Virtual Data Center (VDC), which will be charged as a monthly service fee from a local provider. The VDC platform offers scalability and change options, since hardware and the scope of the project are subject to rapid technical advances and growth. The VDC concept also provides the most efficient and secure avenue for multiple agency access to ALPR data.
The SOSINK project is currently in the planning and early deployment stages, with the entire system scheduled to be operational later this year.
Implementation Considerations and Lessons Learned
As ALPR technology is experiencing rapid growth (or deployment) within agencies around the world, it is useful for agencies to share implementation considerations and lessons learned.
First and foremost, determining the goals of the program and the type of ALPR system best suited to accomplish these goals is paramount to the program’s success. Is the system designed for tactical coverage of trouble spots or special events, surveillance of key access points into an area, broad coverage of a region, and/or the collection of intelligence? Depending on the goals of the system, the appropriate mix of fixed, mobile, and portable systems must be considered.
Next, the capabilities of various vendors must be reviewed thoroughly to ensure alignment with an agency’s priorities. A review of other agencies’ programs and an evaluation of vendor capabilities must be included in the project planning. In addition to cost, considerations to be weighed in vendor evaluations include the depth of the offering, support capabilities and responsiveness, recent financial performance, number of engineering and support personnel available, system accuracy as reported from an independent study, length of time in the ALPR business, number and quality of references, software release schedule, and OCR capabilities.
Agencies should also be able to get answers from vendors on the following questions. Does the vendor design and manufacture all components of the system? Does it have total engineering control of the system, or are some of the components purchased elsewhere and integrated into its solution? A vendor with control of its own proprietary system is preferable, as it will be better able to respond to an agency’s changing needs and/or provide support.
Where is the vendor headquartered, and what resources are available to client agencies? Does the vendor have a parent company? If so, where is it headquartered, and is it committed to the business? What is the financial strength of both the vendor and the parent? It is important to know that a vendor is financially viable and will be there to support an investment over time.
How long has the vendor been in the ALPR business, and how many references can it provide? More importantly than the number of references, what is the quality of the references provided? A list of pilot or demonstration locations or a list of agencies with no more than one or two systems deployed should raise concern. Does the vendor listen to its customers and innovate its product based on this feedback? What is that process, and how frequently does the vendor update its product?
The importance of the OCR capability cannot be overstated. Regardless of how good the cameras, software, and other components are, if the OCR performance level is not adequate, the value of the resulting data is minimized. OCR plays a critical role in determining the system’s accuracy. With the license plate environment in the United States continually changing, a dedication to OCR is essential. Does the vendor maintain dedicated resources for OCR refinement? How many OCR “engines” does it maintain, and does it have (or is it willing to create) an engine that will provide optimal performance in a given area? The CPD worked closely with its vendor, Federal Signal’s PIPS Technology, to develop a custom OCR engine for the Cincinnati region. Given that Cincinnati is located in a tristate area, it was important to optimize the OCR engine to Ohio, Kentucky, and Indiana plates. The collaborative effort also improved the accuracy of vanity plate results by removing extra characters and symbols such as the C-shaped paw for University of Cincinnati Bearcats plates or the B for Cincinnati Bengals plates.
System accuracy is a function of how many plates are captured and how accurately those captured plates are interpreted. A system that captures every plate it encounters but inaccurately interprets those plates is useless. Likewise, a system with very high OCR interpretation accuracy that captures only a small percentage of plates is also of little value. Good systems will have technologies in place that optimize both sides of the accuracy equation: capture and interpretation. Many agencies will report on the accuracy that they have seen, but many of these evaluations are subjective and/or biased by relationships with vendors.
Once an agency has selected an ALPR vendor, a project plan should be put together to facilitate deployment. This plan should include elements of how and where the technology will be deployed for the greatest benefit as well as how the officers using the system will receive training and ongoing support.
For mobile systems, agencies should consider the issues of take-home vehicles, repetitive beats, officer driving patterns, and individual officers’ abilities to learn and use new technology. Investing thousands of dollars in an ALPR system that will sit in someone's driveway much of the time, to be used only on the same beat every day, is a waste. Similarly, installing ALPR technology in the vehicle of an officer who is unable to respond to the hits from the system because of calls for service is probably not the best use of this technology. The CPD has found the greatest success in assigning officers with ALPR-equipped patrol cars to a specific geographical area but with flexibility in patrol responsibilities. ALPR-equipped vehicles are also required to be deployed a minimum of two tours of duty per day. Understanding and mitigating issues related to use will enable an agency to get the best return on its investment.
CPD officers are required to attend in-depth training on the technology, which includes one hour of classroom and four hours of in-car instruction with department-certified instructors trained by the vendor at its headquarters. Training includes topics related to hardware, software, support, and future development efforts. Newly trained officers also ride with an experienced ALPR officer for multiple shifts to ensure their ability to use the system successfully. The CPD has learned that it is neither ideal nor necessary to train every officer. Identifying officers who will use the system to its fullest potential is the best approach.
To support the ongoing use and effectiveness of the system, a strong information technology (IT) foundation is essential for user and hotlist management as well as for server and storage capabilities. Who will be the system’s users, and what access privileges will need to be defined within BOSS for the different groups of users (command staff, officers, investigators, and other agencies accessing data remotely)? Agencies should identify relevant hotlist data sources and identify their contacts for obtaining access to NCIC, state, local, and other hotlists. Consider how often hotlists need to be updated and customized, and define this within BOSS. Do all hotlists go to all mobile and fixed cameras, or are certain hotlists specific to certain mobile units? By what means are the data to be communicated between BOSS and the mobile and/or fixed cameras (wireless, thumb drive, cellular, wi-fi, fiber)? How are hit notifications to be managed—are they given to the officer in the vehicle, are they sent to a central dispatch location for review, or are they sent covertly to an e-mail address without notifying the officer? Agencies must determine how much data will be stored and for what duration, taking care to consider any state or local requirements for data retention. These are all items to be managed by IT staff as an administrative function within BOSS.
A Double Homicide Solved
In the summer of 2008, the Cincinnati Fire Department was called to a report of a fire at an apartment building. Upon arrival, it was discovered the fire was set deliberately to cover and destroy evidence of a double homicide of a prominent retired doctor and a well-known local hairdresser and makeup artist. As the CPD began to process the scene, much of which was damaged, a relative of one of the victims noticed the victim’s vehicle was missing. After confirming the information, a BOLO notice was sent out to the entire county. A CPD officer operating an ALPR unit at the time recalled possibly seeing a vehicle matching the description an hour or so earlier. He immediately returned to the district and loaded the images from his vehicle into the BOSS database. Within minutes, investigators had the vehicle’s image, which was printed and distributed to all available officers, as well as its recent location. Shortly after officers began canvassing the area, the vehicle was located with two occupants. This was the lead and the break investigators needed. By quickly locating this vehicle, key evidence was preserved, giving investigators valuable information they likely would have lost had the vehicle been located hours or even days later. With this lead and additional investigatory work, Anthony Wayne Scott was charged with 11 counts including aggravated murder, aggravated arson, murder, aggravated burglary, and aggravated robbery. He is currently awaiting trial. In this case, the ALPR system, coupled with good old-fashioned police work, has clearly helped in taking a very dangerous man off the streets. Any agency considering ALPR systems should look at this case to help justify the value of such a system. Although it is easy to place a value on recovered fines and stolen vehicles, it is important to consider the value of staff hours saved through more efficient investigations and the value of promptly locating and preserving evidence.
ALPR technology has had a significant impact on the CPD’s ability to protect and serve the residents of Cincinnati. The prior successes and the planned expansion and regionalization of the system clearly illustrate Cincinnati’s commitment to its ALPR program. Once deployed and operational, the SOSINK system will likely be one of the most advanced and well-planned ALPR networks in the United States and should serve as a model for other agencies and regions looking to deploy this technology. ■