Thirty Years of ALPR

Police car

Three decades after its invention, where does ALPR technology stand and what do you need for effective implementation?

Since Automated License Plate Reader (ALPR) systems first gained recognition as “force multipliers” for North American law enforcement, they’ve enabled significant improvements in everything from stolen vehicle recoveries to criminal intelligence gathering.

As a 2014 RAND report noted, however, expiring grant funds, privacy related criticism and a lack of research are all causing police administrators to reevaluate ALPRs which compete for other high-tech tools, such as body-worn cameras, digital forensic analysis tools, robots, drones, facial recognition, and others.

In addition, data retention concerns; logistical issues such as the installation, maintenance, and repair of ALPR hardware; and interoperability needs with neighboring agencies’ ALPR systems can all affect deployment.

These issues are far from insurmountable. Instead, what’s needed is a solid foundation not only in ALPR’s technical capabilities, but also in how they and other technologies relate to your agency’s mandate. Understand this and you can understand – and, more importantly, communicate – its strategic value to your agency and community.

ALPR: Where It’s Been and Where It’s Going

An ALPR system’s core components remain the same now as when it was first invented: high-speed cameras; Optical Character Recognition (OCR); a way to compare the characters rendered from OCR to databases or “hot lists” of license plates of interest; and an interface which alerts police officers when there’s a “hit” on a hot list.

Saving Costs with Commercial Hardware

How those elements come together has evolved over the years. For example, no longer is infrared a strict requirement. “The advent of the megapixel IP camera using white LED light allows newer LPR systems to use color and high definition without need for infrared,” says Steve Lewis, Business Development Executive for Michigan’s OpenALPR Technology. “Users can also obtain more forensic details; not only the tag itself, but also the vehicle’s make, model and any identifying marks.” model and any identifying marks.”

Such high-definition cameras are no longer limited to proprietary hardware. Now, they’re available off-the-shelf in commercial models. In addition to better detail, they allow plate capture from a wider field of view – for example, all lanes of a superhighway – longer distances and different angles, as well as the capture of multiple plates from one image.

Even mobile device cameras can be used for ALPR. Mobile apps, says Tom Joyce, Vice President of Business Development for California-based Vigilant Solutions, offer value in situations such as the need for a “vehicle canvass” during response to a major incident.

In these and other situations, ALPR removes challenges associated with handwritten notes, including illegible writing and transposed characters. In turn, ALPR adds the value of metadata – the date, time, location, and any additional images of associated vehicles.

Thus, not only can cameras – mobile, portable, fixed, or microsized – be configured to fit whatever environment they’re deployed in, agencies can now opt to deploy software only solutions such as Vigilant’s, OpenALPR, or PlateSmart, driving down costs from the tens of thousands into the thousands – or even less. OpenALPR’s software is open source, meaning that organizations can download and use it at its most basic level for free. For additional features such as high accuracy US state of origin details, H.264 advanced video coding support, plate grouping, high-speed processing, and access via a Web app, agencies can pay by the month. (Lewis says some agencies have used this option to run temporary operations. In one example, an agency used the technology to automate stakeouts. Rather than pay agents to monitor the homes of fugitives and their known associates, ALPR technology scanned for the fugitives’ plates and alerted agents when they got a hit.)

No Data Collection? No Problem

In fact, agencies no longer even require ALPR hardware to have an ALPR strategy. OpenALPR is commonly downloaded and deployed by private companies, such as HomeOwner’s Associations (HOAs) and property managers, for use with their own surveillance systems. This, Lewis says, augments traditional video surveillance, allowing users faster access to the context of what was happening around the area where the vehicle appeared.

Companies using ALPR this way routinely notify law enforcement of suspicious vehicles. In one example, a Denver area HOA placed a fixed ALPR camera at its community entrance. When the driver of a strange vehicle tried to abduct a young girl, the camera helped to identify a vehicle which matched the description. Once notified, police were able to apprehend the suspect.

A St. Louis, Missouri-based company, Watchtower Security, has taken that model one step further. The managed video surveillance services provider partnered with OpenALPR in September 2015 to improve its forensic video analysis services. “Watchtower cooperates with local law enforcement by providing timely information (plate numbers, vehicle color and make and model) to speed investigations,” says Lewis.

A more historical approach is possible with Vigilant which owns its own database of images populated by its member agencies and commercial resources. Vigilant allows analysts and investigators to use its software to access its online environment. From there, they can compare data and receive E-mail or push alerts on their own hot lists, or preconfigured hot lists from, for example, the NCIC stolen vehicle database.

Multiple Data Storage Options

Agencies don’t need to rely on third parties for storage, however. Many have fed a demand for cloud-based storage to maintain not just ALPR data, but all digital evidence collected from incar and body-worn camera systems, as well as other systems.

As a result, many ALPR systems integrate with cloud solution providers, such as Microsoft® Azure® Government or Amazon Web Services (AWS), both of which implement best practices to meet standards which can be considered compliant with – either meeting or exceeding – Criminal Justice Information Services (CJIS) requirements and others, such as the International Organization for Standardization (ISO).

Agencies which can’t budget for a larger-scale cloud storage solution might opt for a “hybrid” like OpenALPR’s Cloud Stream solution in which an agent running on a local server or laptop sends metadata to the cloud for displaying results. The processed image data is then retained locally, depending on the agency’s storage capacity and retention requirements. (Lewis says OpenALPR will introduce its own ALPRaaS, or ALPR as a Service, including cloud processing and storage later this year.)

New Information Sharing Opportunities

These advancements have all helped to facilitate regional efforts to consolidate data and cooperate crossjurisdictionally in  initiatives as complex as the Automated Regional Justice Information System (ARJIS) of Southern California which relies on compliance with the National Information Exchange Model (NIEM) to share ALPR and many other types of data.

Information sharing isn’t only a matter of the right technology integrations, however. Joyce says Vigilant’s goal is to expand beyond its existing 1,000 customers to create a nationwide sharing environment, providing agencies in one state with the technology to share the ALPR data they collect with agencies in any other state – provided the appropriate memorandums of understanding (MOUs) are in place. “It would be the technical equivalent of accepting a Facebook friend request,” says Joyce. While a onetime custom integration of external camera data might be necessary for agencies using competing hardware, the end goal would be ingesting the data into Vigilant’s LEARN database.

From Collection to Analytics

The ability to filter these ever increasing data sets is of growing importance to law enforcement. Vigilant recently introduced a filtering capability by vehicle year, make and model. “Taking all scans in a quarter mile radius over a 24 hour period might yield three thousand vehicles,” says Joyce. “By filtering that list through an eyewitness description of a suspect vehicle, you can reduce that list to maybe a dozen leads as a starting point.”

Lewis agrees, saying that much of ALPR’s resurgence comes from “the Big Data movement,” and machine learning that makes LPR more accurate overall. “You can feed the system with tens of thousands of vehicle images and data points,” he says, “so that it can learn what, for instance, Toyota® Camrys look like. Then, as it processes new images, it can find suspect vehicles faster than with license plate characters alone. This relieves witnesses of the responsibility of having to recall full plate numbers and other details.”

These aren’t the only analytic capabilities. Vigilant includes GPS data in their plate collections which can help to identify vehicles seen in and around crime scenes. This gives investigators a mechanism to predict “the most logical place and time to look for a vehicle based on historical scans,” says Joyce, taking ALPR from reactive alerting to proactive analysis and, even, prediction.

While Lewis isn’t aware of any ALPR systems which integrate with crime mapping or predictive analytic software, he says OpenALPR’s API would allow developers to create their own plugins for this level of functionality, embedding the OpenALPR libraries algorithm within other software solutions.

Over time, the addition of other data, including biometrics, may become possible. Joyce says that Vigilant, having also introduced its FaceSearch technology – the aggregation of mug shots from multiple sources such as sex offender registries and CrimeStoppers Web sites – is exploring integration with its ALPR data in ways which don’t violate anyone’s privacy.

The Privacy Implications of ALPR Data

Privacy is a key sensitivity. Much of the attention on digital data, such as evidence found on mobile devices or body-worn cameras, focuses on collection of that data. With ALPR data, the problem isn’t collection; license plates and vehicle details in plain view on or from a public street have no expectation of privacy.

To that end, ALPRs may not seem like they carry the same privacy or personal liability risks which body-worn cameras do. David Roberts, Senior Project Manager with the International Association of Chiefs of Police (IACP), explains that a body-worn camera might bring its wearer into a private residence late at night, as well as being activated at moments of personal confrontation. ALPRs, in contrast, collect data passively and in public.

Roberts notes that many agencies’ ALPR usages started with small goals, like recovering stolen vehicles, and then quickly expanded as more investigators realized the potential for other kinds of investigations. Under these circumstances, the apparent need to retain data beyond just a few hours or days became more critical.

Still, the volume of data collected presents concerns around the “mosaic” of personal information which could be reconstructed from indefinitely stored ALPR data and metadata such as GPS coordinates. That’s why the retention of ALPR data sparks privacy concerns more than its collection. The data has undeniable investigative and intelligence value, but is that value indefinite?

Vigilant’s Joyce, once assigned to the New York City Police Department’s Cold Case Squad, recalls decades old homicide cases which point to data being indefinitely valuable. However, he acknowledges that most criminal cases have an average statute of limitations of between five and seven years.

Who retains the data is another concern. Locally retained data – saved on servers within a police department or municipality – or data saved to a cloud service paid for by the agency is not as much of a concern for some, as data retained by a third party which offers the data to agencies, as Vigilant does, for a fee.

While there is some precedent for law enforcement access to third-party records, resources like LexisNexis® Accurint® or TransUnion’s TLOxp still consist of aggregated public records – not citizens’ movements throughout a given region.

The location data attached to license plate snapshots was enough for Roberts and coauthor Meghann Casanova, in the 2012 National Institute of Justice report “Automated License Plate Recognition (ALPR) Systems: Policy and Operational Guidance for Law Enforcement,” to recall a Supreme Court case decided that same year. United States v. Jones (132 S. Ct. 945, 565 U.S. __ [2012]) highlighted the issue in a concurring opinion which took issue of the length of time of GPS tracking – four weeks – and the kinds of private, personal movements tracked together with criminal activity.

While the point in time ALPR snapshots are not nearly as comprehensive as the kind of monitoring allowed by GPS tracking, indefinitely stored ALPR data touches on the opinion’s concern regarding technology with the capacity to “make long-term monitoring relatively easy and cheap.”

Joyce emphasizes that even in Vigilant’s commercial database, ALPR data is never matched to other records including names, dates of birth, addresses, drivers’ license numbers, or other Personally Identifying Information (PII) from public records or state motor vehicle databases. “Knowing generally where a vehicle has been located doesn’t suggest specifically where its operator visited,” he says.

Moreover, he states, he has never seen disparate databases tied together in a way which profiles an individual based on his or her activity. “Most investigative information isn’t used proactively to establish profiles,” he says. “There simply isn’t the time for it.” Cases where vehicular patterns of behavior might suggest, or even confirm, criminal behavior, where such proactive information is valuable, aren’t common enough to warrant concern over everyday use.

That, Roberts says, is where a strong policy can step in. He says he’s encouraged by the forward thinking response to policies and procedures around the use of body-worn cameras. “It seems like there are corollaries between lessons learned from ALPRs and from body-worn technology,” he says, “especially around reasonable usage expectations and how and who should manage, retain and access data.”

Putting It All Together

Policy mixed together with good strategy and training is well worth the investment, adds Roberts, considering the advancements in ALPR which stand to improve its force multiplying attributes.

He stresses that the evolutions in technology tend to expose gaps in both policy and legal precedent. While the courts struggle to keep up with these evolutions and to apply the law accordingly, law enforcement administrators need to be able to demonstrate an ALPR policy – or any technology policy – which aligns with their agency’s strategic objectives.

Many agencies, for example, collect, but then do nothing with the data. This may be because their ALPR “pings” on too many hot lists, overwhelming individual officers who cannot handle all the alerts to handle expired plates, lack of inspection stickers and other minor offenses – even when they get mixed in with alerts for active warrants or vehicles connected to crimes.

This can be a PR disaster, says Roberts, referring to the 2013 suspension of the Boston Police Department’s ALPR program after news media questioned the agency’s ability to act on alerts and to protect the data they were collecting.

“You have to be able to link the data you collect to a business objective,” says Roberts. “This should be based on empirical research showing not just the number of hits to a hot list, but also how many of those were actionable and over what period of time they remain actionable. If the data isn’t that valuable past 60 days, then that’s a reasonable benchmark for your agency’s retention period.”

Some ALPR metrics which Roberts recommends as “universal” include the proportion of captures, read accuracy and matching efficiency, as well as the number of vehicle recoveries. Measuring these numbers over time can help chart a course for evaluation during the procurement process, either for renewal or when evaluating new technology from a different vendor.

This way, in spite of a dearth of research on contemporary ALPR units to indicate whether they are improving, law enforcement administrators can still have a way to anticipate inevitable changes. “You must constantly evaluate technology in light of your business objectives,” says Roberts.

Moreover, the technology should be able to support those objectives, including, for example, database technology which can allow for scheduling an automated dump of irrelevant data on an interval defined by the agency’s goals, strategy and policy. Vendors should, in turn, be able to implement data storage requirements around storage, security and other needs.

All of these factors can figure into your ability to communicate transparently with city councils and other community groups which are stakeholders in defining business objectives and green lighting technology which helps to reach them. Because the evaluation process can be as much of a constant as the rate of technological change itself, law enforcement must take the lead in providing information and correcting misinformation.

Fortunately, the path to better policy doesn’t mean having to start fresh with ALPR or any other technology program. “Communities of practitioners can serve as resources for others,” says Roberts, pointing to organizations like the IACP and state chiefs of police associations, the Police Executive Research Forum (PERF), the National Sheriffs’ Association (NSA), and others, as well as the Bureau of Justice Statistics, the National Institute of Justice and other government bodies which offer model policies and toolkits.

These tools, such as the IACP Technology Policy Framework, can help offer structure when rapid technological changes make it difficult to keep up. Younger officers and administrators can also help smooth the transitions an agency needs to make when technology evolves, such as helping to train other officers in operational practices like verifying the system works and is properly configured before each shift begins. “A policy which is a work in progress is better than having none at all,” says Roberts.

ALPR has made many strides over its 30 year lifespan and is expected to make many more in the future. By maintaining an ongoing dialogue with both vendors and peers, law enforcement commanders can understand how the technology continues to shape and be shaped by policing – and how best to use it to lead their agencies further into the millennium.

Christa M. Miller is a freelance writer and consultant based in South Carolina. Since 2001, she has been writing about public safety issues, ranging from policy and procedure to the use of high tech in law enforcement.