If you've ever worked patrol duties, this scenario should sound familiar. You re patrolling Anytown, USA, in heavy holiday traffic when you spot a vehicle traveling alongside you that triggers that little spot on the back of your neck. You angle your head in a near-impossible position to read the license plate, keeping one eye on the road while your hand makes a furtive grab for the microphone. As you radio dispatch, you intently watch the actions of the driver: Are they looking straight ahead and not at you? Are the passengers stealing glances at you? Is that a screw missing from the rear license plate?
You feel that welcome adrenaline in your veins as you realize you may have stumbled onto an occupied stolen car. But, to be safe and sure, you want to confirm your catch with dispatch. Unfortunately, the radio is tied up because someone else has stopped out for a stranded motorist and has decided to relay 16 possible phone numbers for dispatch to try. So, you decide to try plan B, which is to ignore all your department policies about operating a MDT while the vehicle is in motion, and start frantically pecking at the keys like a chicken at a carnival sideshow. Of course, the entire time this is going on, the occupants of your suspect vehicle get wise to your antics, make an abrupt right turn off the roadway and disappear into a shopping center, blending in with the zillions of shoppers in the lot. Disappointment and frustration replace your former excitement as you resign yourself to losing another one to bad timing, an overwhelmed radio system and lack of crucial information.
Now, consider another scenario. You re cruising along in that same traffic, thinking about last night s episode of The Simpsons, when your MDT begins to squawk like bad karaoke in a Bakersfield bar. You re jolted to reality and look over to see your MDT s screen flashing. The screen displays two small camera shots, one showing the vehicles on your immediate left, and the other showing the ones on the right. Beneath each is a close-up of the license plate of the vehicle captured on the screen. The screen on the right is flashing rapidly. A check into the vehicles of interest list on your MDT determines that this particular plate belongs to a suspect wanted in a series of home-invasion robberies. Your MDT automatically sent an alert message to dispatch and other patrol units in the area when you registered the hit, so the radio channel is locked-down for you, and other units take advantage of the GPS feature to immediately find you. Now having proper backup and priority communication, you execute a felony traffic stop and take the suspect into custody without incident.
The difference in the scenarios? The inclusion of license-plate reading technology into your daily patrol environment. While the technology may sound like something out of Star Wars, the reality is that automatic license plate recognition (ALPR) is already a reality in patrol cars across the country, and it s proving very effective in apprehending all types of criminals. To learn more, I contacted PlateScan, one of the leaders in ALPR. The company, based in Newport Beach , Calif., currently has its proprietary PlateScan technology in patrol vehicles in Los Angeles, Houston, Sacramento, Reno, Mesquite (Texas) and other places. The results have been impressive.
According to PlateScan, ALPR technology was developed in response to a number of high-profile terrorist attacks in the UK in the early 1990s. The system, developed by a company called Racal Research, was aimed at preventing the use of roadways in implementing terrorist attacks. More specifically, it was designed to detect the use of vehicle-borne improvised explosive devices before they reached their intended targets. The system was required to work in all weather conditions and visibilities, and at high speeds. The result was a virtual ring of steel around cities like London, with many criminals detected and apprehended through rigorous use of ALPR technology.
Since that time, ALPR has been implemented in more general policing activities, such as identifying stolen vehicles, vehicles associated with felony warrants and missing persons, as well as more mundane uses such as registration and parking violations. It also found its way to the United States.
How It Works
Departments can implement ALPR as a fixed-location system, a mobile system or a command-center configuration. For the purposes of this article, I ll focus on vehicle installations.
According to PlateScan, a typical vehicle installation involves three main components. The first is the PlateScan computer, which departments usually mount inside the trunk area of a patrol vehicle. This is the brain that processes plates for recognition. PlateScan is unique in that it uses technology known as neural networks, which basically means it uses sophisticated algorithms to identify characters on plates. As the system reads a license plate, it compares the plate to the database. If it makes a hit, it alerts the operator. If not, it stores the plate information in the computer, as well as the time and location of the read via GPS integration. According to Plate-Scan, testing of its systems indicated an accuracy rate of approximately 95.8 percent.
Interestingly, PlateScan can also generate hits off of partial plates. For instance, the LAPD was investigating a series of rapes, and in one of the incidents determined a partial plate of three characters. Using the system, 170,000 stored plates were analyzed in approximately five minutes time, with an analysis of possible plate matches based on characters, time stamps and GPS locations. The result? Six images in full color of a vehicle suspected of involvement in the attacks. In this case, they were looking for a rapist, but you can modify user-defined lists of vehicles of interest to fit your specific needs.
The second part of the equation are multiple, dual-lens cameras mounted on the vehicle. Typically, this involves installing four cameras, two on the roof facing forward, and two on the rear window shelf facing rearward. The cameras used by PlateScan are both infrared (IR) and color for several reasons. First, because many states use reflective paint that reflects IR light on their plates, the entire plate reflects evenly, preventing the system from recognizing fonts and characters by IR alone. Second, where IR can read plate characters, age can make them less reflective, making them harder to read. Third, the application of non-reflective coatings to license plates (remember those ads for beating red-light cameras at intersections?) can make them invisible to IR. By using color cameras in conjunction with IR, PlateScan says these problems are eliminated. It also means that while an IR camera only recognizes the plate due to the reflectivity, the addition of a separate color camera allows the capture of vehicle description, make, model, color, damage, unique characteristics, etc. to the database. In other words, you get the complete picture of the vehicle, not just the plate. The PlateScan system can read and process four cameras simultaneously, which allows for multi-directional, real-time analysis.
The third component is the integration of the system into the vehicle s MDT, MCT or PlateScan-supplied touchscreen. This serves as the interface between the operator and the system. Because the system remains in the background unless it registers a hit, the MDT/MCT functions as normal. This means training time remains fairly short because the system runs on its own unless it needs to notify the operator of a hit. Using the system, the operator can bring up a history of any plate to see if it has been read by any patrol vehicle within a given time frame or location. This proves especially useful when investigating crimes such as residential burglaries. By comparing where and when a vehicle has been read relative to when a burglary occurred, a correlation can potentially exist that can help pinpoint suspected vehicles involved. This is also important when investigators wish to avoid direct contact with a suspect (i.e., surveillance) but need a record of location, time, etc. to establish a case.
While ALPR technology is new to the U.S., it s not new in Britain. The 2001 implementation of ALPR in the county of Northampton, UK, resulted in the following enforcement actions based on 3,501 traffic stops over a 12-month period by only six police officers: Of the 601 arrests, 86 percent were local criminals, 145 were for auto-related crimes, 166 were for other crimes, 200 were for driving with a suspended, revoked or no driver s license, and 90 were wanted persons for various reasons. The arrests also included the recovery of a kidnapped girl, the arrest of her captor, drug arrests and others. More impressively, the arrest rate for these six officers ran nine times the national average for police officers in the UK.
Closer to home, the installation of PlateScan (a direct descendant of the system discussed in the British study) in one of the LAPD s patrol units allowed the scanning, recognition and processing of more than 1,000 cars in its first hour of use. Moreover, it allowed the detection of a stolen vehicle, and ultimate arrest of its driver.
So, it seems that ALPR works. As for cost per unit, it varies based on whether existing hardware is used, so PlateScan constructs and prices the system on a case-by-case basis. (You can implement the PlateScan ALPR system on an existing MCT set-up, and it can use existing DVR camera setups, thereby reducing overall cost and installation challenges.) PlateScan states that a complete system from them runs approximately $20,000 per car, which can vary depending on the number and type of cameras used. The bottom line: Crimes solved through use of PlateScan and other ALPR systems produce a benefit much larger than any dollar amount. For those with more fiscal viewpoints, ALPR can recoup thousands of dollars in unpaid parking and registration violations, easily justifying the initial cost issues. And for those of us working the street, it can make patrol work a lot more fun.
This article focused primarily on PlateScan, but several companies produce license-plate scanning products.
AutoVu Technologies Incorporated
PlateScan (A division of Civica Software)
Remington Elsag Law Enforcement Systems, L.L.C.
State Trooper JP Molnar has been teaching EVOC since 1991 for various agencies. He has also been racing cars for 24-plus years, and has taught at numerous high-performance racing schools.