What is Automated Number Plate Recognition (ANPR)?

The number of horseless carriages on the roads is increasing every day, and with every vehicle comes a registration number, license, and a couple of drivers. So, for each automobile on the road, a lot of data is processed.

It is hard to monitor and manage this amount of data. Global law enforcement required a solution to help them with this tedious task. Police Scientific Development Branch in Britain came up with a solution, and that solution is ANPR.

What is ANPR?

ANPR (Automated Number Plate Recognition) is a computer vision tool used to read vehicle registration plates. Law enforcement around the globe widely uses it to verify vehicles’ registration, license, identify stolen cars, and monitor traffic. ANPR uses video surveillance as well as traffic enforcement cameras to fulfill this purpose.

Tools and Techniques behind ANPR

ANPR mainly uses OCR to recognize registration plate characters. It captures high-res images of vehicles using specially-made speed cameras and recognizes the plates following a sequence of algorithms.

ANPR takes in a video surveillance feed and monitors it frame by frame. Infra-red cameras with narrow wavelength retro-reflective techniques capture any frames with clear or processible images. The system uses localization to find and isolate registration plates from captured images.

Once a plate is recognized, the tool takes the localized part of images and applies normalization. Normalization adjusts the cropped images into the required dimension, size, brightness, and contrast. ANPR uses the output of normalization and processes it using character segmentation to identify individual characters.

Segmented images pass on to OCR for recognition. This step is crucial as OCR must be able to detect different font faces. High-speed processors aid character recognition and identify if a number plate exists in the system.

Additionally, ANPR uses geo-analysis on recognized images depending on image specifications and the country that uses it. Each capture gives data about a new vehicle or information about an existing vehicle.

Capture

Localize

Normalize

Segment

Segmentation

From top to bottom: Capture, Localization, Normalization, Segmentation, Input for Recognition

How to Create ANPR with YOLO

Let us understand how we can combine multiple technologies to create a small ANPR. We will combine three tools: YOLO, OpenCV, and Google’s Tesseract.

First, we need to install these tools in the system. YOLO and OpenCV can install with pip. To install Tesseract, we need to download and run the binary files available on its website. OpenCV is a tool that we use to capture and process live feed and help in detection.

Once we install the tools, we set up OpenCV on a live feed and apply YOLO on top of OpenCV to detect moving vehicles. YOLO identifies vehicles in fast-moving traffic. Whenever a car passes in the video stream, YOLO identifies it and captures the frame. We apply the given image processing algorithms and extract characters from the capture.

The capture gets passed to the Tesseract that processes to recognize characters. Characters identify as an array which we concatenate into a string.

Applications of ANPR

Applications of ANPR mainly focus on law enforcement and traffic crime prevention but are not limited to these. The private sector uses ANPR for security and management. Government agencies globally use it for the automation of many traffic-related processes.

ANPR uses include parking management systems, journey time analysis, weighbridges, tolling, bus-lane enforcement, and intelligent transport systems.

Wrapping Up!

We hope you find this article helpful. ANPR is an effective tool as well as a fusion of tons of great technologies. It carries the legacy of decades-old efforts of scientists and enlightens us on how quickly technology is evolving. It contains a mix of state-of-the-art hardware and top-of-the-line software.

Happy Learning.