Real-time ANPR on live CCTV streams.

Security & Surveillance / Computer Vision / OCR

The project focuses on developing an algorithm capable of recognising vehicle number plates from images, recorded videos and live video streams. The system is designed to work fast enough to detect the plates of fast-moving vehicles, and it creates a record/log of vehicle entries and exits — making it useful for automated vehicle monitoring and access-control applications.

Business requirements & challenges

Business requirements & challenges

The system needed to read number plates reliably across multiple input types, fast enough for moving traffic, and maintain an auditable record of vehicle movements.

Key requirements

• Recognise vehicle number plates from a given image, video and live video streaming
• Be fast enough to detect numbers from fast-moving vehicles
• Create a log of vehicle entry and exit for monitored locations

Solution

Solution

The solution uses a sequence of vehicle detection, number-plate localisation, image enhancement, character segmentation and OCR.

• Vehicle detection from image, video or live video stream
• Number-plate localisation via a dedicated segmentation algorithm
• Plate enhancement so further processing can be performed effectively
• Thresholding and character segmentation to separate individual characters
• Optical Character Recognition (OCR) applied to each character to extract the plate number

Stack

Technologies used

How it works

Process workflow

1. 01Video capturing from CCTV — captures video of vehicles entering or passing the monitored area (also works with images or streams)
2. 02Automatic detection of number plate — detect the vehicle and localise its plate via segmentation
3. 03Text extraction from number plate — enhance the plate, then threshold and segment the characters
4. 04Number-plate text recognition — apply OCR to the individual characters to extract the plate text
5. 05Verify from number-plate database — check the recognised plate against a database (registered / authorised / expected / potentially unauthorised)

Where it's used

Applications

Business benefits

Business benefits

• Supports multiple applications — vehicle records, tracking, rule-violation detection, penalty billing, toll and parking-fee collection
• Increased revenue — automatic detection of rule violations without relying entirely on manual monitoring
• Prevents trespassing — identifies vehicles at private-premises entry gates to stop unauthorised entry

At a glance

Key highlights