License plate recognition (LPR) technology is undoubtedly one of the best tools to come along in a very long time. More than any other equipment that I can think of, LPR is both an effective crook catcher and a valuable force multiplier. As with any new tool, LPR has its share of misconceptions and outright misinformation circulating. When you’re making procurement or deployment decisions, you need solid information, so this month we’re shining the light on nine of the most common myths associated with LPR.
No 1: My Vendor Is Best
It’s human nature to be competitive; we all want to believe that what we have is better than what the other guy has. You’re certainly not going to hear anyone say, “Yep, we did a lot of research and decided to go with the second-best brand.” I’ve heard time and again from different agencies that the brand they’re using is the best. Reality check: There’s more than one great LPR vendor (see “LPR Heavy Hitters,” May 2009). This is a relatively new technology for law enforcement, and we’re early in the adoption and product development cycle. New features and capabilities are being added regularly, and the market is very competitive right now. My experience is that multiple companies have great products that can deliver effectively.
No. 2: LPRs Are Primarily for Finding Stolen Cars
It’s true that most agencies acquire LPR technology with the primary idea of addressing auto theft and finding stolen cars. And it’s true that LPR does a great job of identifying stolen vehicles (see “Stolen Car King,” March 2009, for a story on an Arizona Department of Public Safety trooper who has identified more than 400 occupied stolen cars). However, the real value of LPR goes far beyond finding stolen cars. Once an agency begins taking a look at the LPR capture data, it quickly realizes there’s huge potential for solving crimes and preventing problems. Time and again, major crimes (including homicides) have been solved because of an LPR capture (not a hit) that occurred before the crime was even discovered. For a great story on this, take a look at “Search the Data & Solve the Crime,” June 2009.
No. 3: LPR Takes the Human Element Out of Policing
I’ve heard this comment more than a couple of times, and it surprises me every time. There seems to be a certain segment of our profession that’s embracing the stereotype of “100 years of tradition, unimpeded by progress.” Yes, most officers drive around their beats and do a pretty good job at identifying the bad guy. But LPR takes this capability to a whole new level. Officers are human, and they can look only one direction at a time and can remember only a limited amount of information. Plus, they have other things going on, like driving, monitoring radios and looking for bad guys in general. With LPR, a good officer becomes a great officer because the technology extends the officer’s capabilities significantly. And, bottom line, the LPR is only providing information; it’s up to the officer or investigator to act on that information.
No. 4: LPR Is Too Expensive for Small Agencies
The entry point for getting operational with a multi-camera mobile LPR system is about $20,000-plus, and it can be higher if you don’t already have a computer display in your vehicle. Given the price and the current economic climate, it’s understandable that many agencies are inclined to pass, especially if they don’t have a thorough understanding of the capabilities. But the reality is that this is exactly what you need when your dollars are tight. LPR is the ultimate force multiplier. Once operational, LPR allows officers to be much more effective and solve more crimes. I can give you several real-world examples of homicide cases that were quickly brought to closure due solely to LPR use. One of them involves the killing of five family members in their home. Major cases tie up huge resources and burn through dollars very quickly, not to mention how they cause credibility problems with the public. Time and again, agencies have realized a quick return on their investment in LPR.
Speaking of investment, there have been agencies that have acquired LPR with the idea of identifying parking scofflaws. Yes, I know, this isn’t nearly as exciting as finding homicide suspects, but it’s a reality of policing. Once again, LPR has been very successful at this task. Many agencies have quickly paid for their systems with the parking fines recovered as a result of towing or booting cars that have significant levels of unpaid tickets. And the side benefit is that the LPR read captured by the parking officer today may solve a homicide or robbery case tomorrow.
No. 5: The Accuracy Rate Is THIS BIG!
One of the most common questions about LPR cameras concerns their level of accuracy. Some vendors are often quick to throw out a number (usually in the mid-90th percentile) that may or may not truly represent the equipment’s capability. There are a lot of variables (besides the vendor) in reading license plates effectively, including weather, distance, speed, type and condition of plate, state of origin, angle of capture, partial obstructions and so on. Understandably, most vendors want to present their product in the most favorable light possible, and sometimes the figures they provide can be a little misleading.
There are two general approaches to determining accuracy levels of LPRs. The first is the approach vendors take. The second is a more realistic and objective approach that I strongly recommend any agency use to compare LPR equipment.
The typical vendor approach: Conduct a test session and review all data captures, then look at the photo of the plate and compare it to the resulting read provided by the LPR equipment. If the photo shows ABC123 and the LPR reads ABC123, that’s an accurate read. Pretty straightforward, right? Well, here’s the dirty little secret: The systems won’t tell you how many plates they didn’t even try to read. Sometimes cameras won’t detect the presence of a plate and will totally ignore it. For example, if you drive an LPR vehicle through a parking lot with 1,000 cars and the reader picks up on 800 of them, reading 775 of the 800 accurately, then the vendor will report that the accuracy rate is approximately 97% (775 out of 800), and that’s an impressive number. However, what you generally won’t be told is that the reader failed to even recognize 200, or 20%, of the plates. If this number were factored in, the LPR would be 77.5% accurate (775 plates out of 1,000 were captured and accurately read).
The better approach: Use a parking lot containing a known number of vehicles. If possible, have a good mix of plate types (including some out of state) and a mixture of vehicle types so the plates are being read at different heights and with real-world challenges, like trailer hitches. Drive through the lot with each vendor and in a manner as consistent as possible. Upon completion, check the number of captures and then go back and examine each read to see if it was accurate. This takes a while, but it’s important. Now, look at your numbers, and you’ll have a much better idea of what your real-world experience is likely to be. To give you an idea of how important this is, let’s use the example of 1,000 vehicles in the parking lot again. If Vendor A reads 950 plates out of 1,000 and gets 925 of them right, then that’s an overall capture and accuracy rate of 92.5% (925 out of 1,000). If Vendor B reads 700 plates and gets 690 of them right, the capture and accuracy rate will be 69%. Note: Under the typical vendor approach, Vendor B would likely claim an accuracy level of almost 99% (based on 690 out of 700 read)—even though it failed to read 30% of the vehicles in the lot!
No. 6: Ours Read All 50 States
I once got into a significant discussion with an LPR company vice president who strongly implied that theirs was the only camera capable of reading plates from any of the 50 states. The reality of current LPR technology is that, for complex reasons, cameras have to be tweaked during their configuration to maximize the effectiveness in the area where they will be deployed. Basically, the combination of fonts and colors present a big challenge to LPR vendors, and it’s truly amazing that the equipment works as well as it does. If a camera is going to be deployed in Arizona, for instance, the equipment will be configured to maximize the effectiveness with that state and possibly the surrounding border states. Plates from farther afield are less likely to occur there, and the system likely won’t work nearly as well when reading them. This is not to say that a read won’t be attempted—just that the capture and accuracy will vary greatly when other states and plate configurations are involved.
When challenged on the 50-state camera claim, the VP eventually acknowledged that even though they claimed 50-state capability, the cameras would have to be individually configured as described above before they could actually read 50 states. In other words, if you were to run plates from all 50 states past their reader, many of them would not be read. This is not an indictment of any particular vendor, just the reality of a limitation in the technology due to the intricacies of infrared wavelengths and the way certain plates use certain color patterns.
No. 7: Digital Camera + OCR = LPR
When LPR technology started gaining momentum, some enterprising companies quickly tried to jump on the bandwagon by offering to match a digital camera and an optical character recognition (OCR) engine with a patrol car’s computer. The results were disappointing. Reading license plates requires very advanced and complex algorithms and is well beyond the typical OCR engines used in scanning paperwork. Traditional OCR is incapable of reliable reads when there are extreme angles, high-speed closure, varied lighting conditions and ever-changing color schemes. Suffice it to say, this is why you want to stick with an LPR vendor who has a proven product and a base of satisfied customers.
No. 8: HOT Lists Are Really Hot
LPR works by checking against a list of wanted vehicles, often referred to as a “hot” list. Because an LPR can check thousands of plates in a shift, the systems work against a downloaded list of plate numbers rather than checking against a “live” database like your state stolen vehicle list. (If the LPR systems were checking live, the state systems would likely crash.) Because of this, the information is only as current as the last time it was downloaded from the state. For a while, most states were providing this update only every 24 hours, but this is changing and many states are now working toward making an hourly update available. Of course, this means that you have to have a way of getting the update out to the car. If your agency relies on a USB update at the beginning of each shift, the update will be only as current as that last upload. This is why officers must verify any LPR hit before they take action.
No 9: Mobile LPR Is Plug-&-Go
Mobile LPR systems can be relatively quick to get up and running. A hard-wired unit can be installed in a few hours, and a magnet-mount unit can be installed in as little as 15 minutes. But this is not quite the end of the story. To realize the full benefit of an LPR system, consideration has to be given to the management of the data and this requires thought and support. Unless you’re going to dedicate a laptop with a large hard drive to a single LPR unit, you probably will want to work out a method of storing and accessing the data that makes sense within your agency’s limitations. For example, investigators will greatly benefit from having access to the LPR data captures, and if the data is riding around inside the LPR car, they won’t have that option. The best approach is to think about this as separate pieces of a system. You need a capture device (the mobile LPR cameras), a data-transfer capability (wireless or USB), a storage ability (large hard drive or server) and a query capability (usually accomplished with the LPR vendor’s support software). Bottom line: This equipment takes a little bit more thought than plug and go. It isn’t a radar unit, and you have to do a little bit of planning to get the most out of your system.
In Sum
LPR technology has incredible potential, and we’re in the early adoption phase. Do your homework when you’re making your procurement decisions, and you won’t be disappointed. At a minimum, check out some of the great articles that we have posted on LawOfficer.com by entering LPR in the keyword search box. You’ll find a trove of great information.