ANPR Systems: How Automatic Number Plate Recognition Works

At factory gates, car parks, rest areas, and logistics warehouses, vehicle entries and exits are still often logged by hand: a guard writes the plate in a book or types it into a computer. It is slow when queues are long, prone to mistakes, and hard to search later.

An ANPR system (Automatic Number Plate Recognition) reads number plates straight from camera images and logs them automatically with the time and a photo. This article covers how it works, the camera requirements, the factors that affect accuracy, and example uses.

How ANPR works

1. Vehicle detection. A computer vision model finds vehicles in the camera image.
2. Plate localisation. The system finds the number plate on each vehicle.
3. Character reading. The letters and numbers on the plate are read, a step often called OCR.
4. Logging and action. The result is stored with the time, camera location, and photo, and can open a barrier, calculate parking duration, or send an alert.

Camera requirements that decide results

• Viewing angle. The camera should face the plate without too steep an angle from the side or from above.
• Plate size in the image. The plate must appear large enough in pixels for characters to be readable. Position the camera where vehicles slow down or stop.
• Shutter speed. Moving vehicles need a fast enough shutter so the plate is not blurred.
• Night lighting. Infrared-capable cameras help read plates at night, including when headlights glare.

Existing CCTV cameras can sometimes be used, but they often need changes to position, lens, or settings to suit ANPR.

Factors affecting accuracy in Indonesia

• Dirty, damaged, or partly covered plates, from accessories or mud.
• Plate variation, including non-standard fonts or colours.
• Weather, such as heavy rain blurring the image.
• Vehicle speed at sites without speed control.

That is why accuracy should be tested at the real site, day and night, not judged from brochure figures alone.

Example uses

• Car parks. Logging entry and exit times automatically, calculating duration, and shortening gate queues.
• Factory and warehouse gates. Opening barriers for registered vehicles and logging visitors or supplier trucks.
• Logistics. Recording truck arrivals, waiting time at loading bays, and departure times.
• Rest areas and commercial sites. Counting vehicles and vehicle types as traffic data for operators.

Integration with other systems

ANPR becomes more valuable when its results connect to other systems: automatic barriers, registered vehicle lists, parking ticket systems, truck scales, or warehouse management. Without integration, ANPR is just a digital guest book. With it, ANPR becomes part of the operational flow.

Privacy and data retention

A number plate linked to an owner or driver can count as personal data. Law No. 27 of 2022 on Personal Data Protection requires operators to be clear about why data is collected, who may access it, and how long it is kept. Set a retention policy, limit access, and put up notices in camera-monitored areas.

Testing accuracy on site

1. Collect real samples over several days, including day, night, and rain.
2. Compare readings with the actual plates in the stored photos.
3. Group the errors: plate not detected, characters misread, or vehicle missed.
4. Fix the main cause, often camera position, lighting, or vehicle speed.
5. Test again after the fix to confirm results improved.

This simple process gives an honest accuracy figure for your own site and shows which fixes matter most.

Measuring operational results

Beyond accuracy, measure the effect on operations: average gate waiting time before and after ANPR, how much manual logging is no longer needed, and how fast vehicle records can be found after an incident. These are usually the figures that convince management most.

Involve security or parking staff from the start. They know local vehicle habits, the busiest hours, and special cases that need manual handling, such as vehicles without plates or temporary plates.

Frequently asked questions

Can ANPR use existing CCTV?

Sometimes, if the angle, resolution, and lighting are adequate. Often the camera needs repositioning or replacement at key points.

Does ANPR work at night?

Yes, with infrared-capable cameras and the right settings.

Where does ANPR processing run?

On a local device at the site or on a server. Local processing reduces the bandwidth needed to send video.

ANPR with IncludeVision

IncludeVision is the AI Vision platform in the INCLUDE ecosystem that turns CCTV into data, including number plate reading and vehicle counting. Processing runs on a device at the site, and results can be combined with other data on the INCLUDE platform. Broader video analytics is covered in CCTV video analytics for plants.