UK How ANPR works
ANPR, or automatic number plate recognition, is a video analytics technology that reads vehicle registration plates from camera images and turns them into text a computer can check. It was developed in the UK in 1976 at the Home Office’s Police Scientific Development Branch and is now used everywhere from supermarket car parks to national road networks. In North America the same technology is called LPR or ALPR.
ANPR cameras are positioned and angled to capture plates clearly as vehicles approach, and infrared illumination makes the reflective plate stand out in any light. The recognition can run in the camera itself or on a nearby computer, which also handles the database lookups and controls barriers, signs or alerts.
-
//
Capture. A camera takes images as the vehicle passes, lit by infrared at night and often triggered by a loop in the road or by video detection.
-
//
Locate. Software finds the plate within the image, even when it is tilted, partly dirty or among other text on the vehicle.
-
//
Read. Optical character recognition, now usually an AI model, turns the plate into text and gives each read a confidence score.
-
//
Act. The system matches the plate and opens a barrier, records a visit, starts a parking session or raises an alert.
ANPR cameras vs CCTV cameras
A general CCTV camera can sometimes read a plate, but a dedicated ANPR camera is built for it. The differences are mostly in the optics and exposure.
| Feature | ANPR camera | General CCTV camera |
|---|---|---|
| Field of view | Narrow, one or two lanes | Wide, the whole scene |
| Shutter speed | Very fast, to freeze moving vehicles | Slower, tuned for a natural image |
| Illumination | Pulsed infrared, typically 850 or 940 nm, synced to the shutter | Visible light or general IR for night viewing |
| Image tuning | Plate contrast first, often monochrome | Colour and overall scene detail |
| Output | Plate text, confidence, overview image, time | Continuous video stream |
| Mounting | Set angle and distance to the lane | Wherever it covers the scene |
Many ANPR systems pair the plate camera with a colour overview camera, so each read comes with a picture of the whole vehicle as evidence.
What affects ANPR accuracy
Read rates of well above 90 percent are typical for a properly installed system, but they depend on installation as much as software. The camera angle to the plate should be shallow, usually within around 30 degrees horizontally and vertically, and the plate should fill enough of the image for the characters to be clear. Vehicle speed sets the shutter speed and frame rate needed. Dirty, damaged, non-standard or deliberately altered plates reduce accuracy, as do heavy rain, snow and low sun.
Plate formats matter too. UK plates use a standard font and retroreflective material set out in British Standard BS AU 145e, which helps ANPR. Systems sold for other countries have to be trained for their fonts, colours and layouts, a point our ANPR vs LPR FAQ covers in more detail.
Where ANPR is used
ANPR has spread from policing into most places where vehicles need to be identified without stopping them.
-
//
Car parks. Ticketless entry and exit, pay-by-plate and overstay enforcement, as part of a car park management system.
-
//
Access control. Barriers and gates at depots, business parks and residential sites open for registered vehicles.
-
//
Policing and enforcement. UK police forces use a national ANPR network to detect vehicles of interest, alongside bus lane and clean air zone enforcement by councils.
-
//
Tolling and congestion charging. Free-flow road charging, such as the Dart Charge at the Dartford Crossing, uses ANPR in place of toll booths.
-
//
Logistics. Yards and distribution centres log trailers and lorries as they arrive, and match them to bookings.
The computer behind ANPR
Some ANPR cameras do everything on board. Larger or more flexible systems use an edge computer in a roadside or car park cabinet, which takes streams from several cameras, runs the recognition, holds the local list of permitted vehicles and drives the barriers and signs through digital outputs or relays. That computer has to work in an outdoor cabinet through summer heat and winter cold, often without fans.
Recognition itself has moved to AI. Modern ANPR engines use deep learning models for both plate detection and character reading, which copes better with angled, dirty and stylised plates than older template-matching OCR. The same computer can then add vehicle make, model and colour recognition, so a read can be checked against the expected vehicle as well as the plate.
The ANPR barrier control case study runs it on compact, fanless PoE computers in outdoor cabinets across UK car parks. Our PoE computers power the cameras from the same computer that reads the plates.
Quick answers on ANPR
ANPR: common questions
What does ANPR stand for?
Automatic number plate recognition. North American equivalents are LPR and ALPR, for (automatic) licence plate recognition.
How does an ANPR camera work at night?
With infrared light. The camera floods the lane with IR, which the retroreflective plate bounces straight back, so the characters stand out even in darkness.
How accurate is ANPR?
Typically well above 90 percent when installed correctly. Camera angle, plate size in the image, lighting and plate condition decide the real figure.
Can ANPR read plates from other countries?
If it is trained for them. Many engines support dozens of countries, but each plate format has to be enabled and tested.
Is ANPR data personal data?
Yes, in the UK and EU. A plate can identify the keeper, so operators need a lawful basis, clear signage and a retention policy.
We supply fanless PoE and edge AI computers for ANPR as standard, with the operating system, camera drivers and I/O configured and tested before they ship to the cabinet.