License Plate Recognition Tutorial

What is License Plate Recognition (LPR)?

LPR is a technology that helps in identifying the license plates of vehicles via processing images. Therefore, LPR is used for both traffic and security applications globally.

Access Control System

How does LPR work?

As the vehicle approaches the gate, the LPR unit automatically ‘reads’ the license plate registration number the access control system. The system compares the registration number with a predefined list and the gate opens only if there’s a match.

LPR Technology highlights

As LPR technology gets popular in security and traffic applications, there are certain highlights everyone should know. For instance, LPR technology assumes that every vehicle has a registration plate for identification. If a car has a unique plate number, you don’t need to install a responder or a transmitter on the car.

Secondly, since LPR is an image processing technology, it relies on an illumination e.g. infrared and a camera to capture the image of either the front or the rear part of the vehicle. After capturing the image, the software analyzes and extract plate information for each vehicle. The extracted data is important in:

· Data collection

· Data enforcement

· Open the gate for an authorized car

· Keep time records for entry and exit (for automatic payment calculations)

 

Thirdly, the LPR system has numerous advantages that include:

· It can be used to fight crime and fraud

· An additional camera can be installed to focus on the face of the driver to improve security

· The technology is convenient for use. Unlike technologies that require a transmitter, LPR does not need to be installed on each car

 

Other LPR names

LPR technology can also be referred by the following terms:

· Car Plate Reader (CPR)

· Automatic Number Plate Recognition (ANPR)

· Car Plate Recognition (CPR)

· Automatic Vehicle Identification (AVI)

· Optical Character Recognition (OCR) for cars

 

Check out the ALPR wiki for more info.

 

Plate and images

The LPR units are based on the images of the plates on the front and/or the rear of the vehicle. If a car is exiting a compound that’s secured with LPR technology, the system will ‘read’ the license plate, compare the license plate number with an existing list and then open the gate if the car is identified/authorized.

However, since license plate numbers of vehicles differ in form, material or shape from one country to another, LPR systems are country-specific. License plates can differ greatly depending on the country or even the state. Some license plates have additional information that’s either attached or written for different purposes. For instance, state identification (in the USA plates – upper smaller row), plate issue date label (USA plates), country flag (Europe –left side), province initials (China –upper row), type of the vehicle (Korea – upper row).

Normally, for most applications, LPR systems do not require this additional information because it is hard to identify it due to resolution, quality, complexity, and variation. However, since the main objective is to capture the unique license plate number of each vehicle, if the technology is installed properly, it is efficient and reliable.

What’s In An Image?

Take a look at the image above. It depicts a front view of a car. In a typical format, it consists of all 256 levels of gray, including 0 (black) and 255 (white).

Also in a typical format, the image will measure 768 x 288 pixels. Each pixel is a picture element – hence its name – and this number of pixels measures 0.2 million elements. This impressive amount of information is processed instantaneously by recognition software, which can read the license plate of the car.

There are a lot of key components of image recognition software that go into accurately reading a license plate. Some of these components include:

Global information, which reads the entire image as a whole

Zoom technology, which accurately locates the license and extracts the smaller details to get registration data

Raw data compression, which takes the end result of the first two elements and compresses it so it can all be viewed as a single image.

Using the image above as an example, the raw image data turns into a single string: W859RKX. The compression rate of raw data into the result string is 1:31600!

Do LPR Systems Really Work?

 

In its earlier iterations, LPR systems had too low of a recognition rate to be really seen as successful. It was impractical to use, especially if there were extraneous circumstances lowering the quality of the image (direct sunlight or faded plates, for example). Recognition software was not quite ready to get an accurate read, at least not in quick enough time.

But recent improvements have made LPR systems far more advanced and reliable. Now they are used in several applications in which license reading is a must-have. Newer LPR systems can also be retrofitted to old systems in order to improve image quality at a reasonable cost.

The recognition capabilities are not always perfect, but errors are always accounted for and updates are made to make the system less flawed. It’s now possible to get trustworthy results on a regular basis.

Elements of LPR Systems

Click on each LPR element to read more

The camera typically takes several pictures of the vehicle, from either the front or the back of the car.  

Every functioning device runs on computer components nowadays. The LPR system is typically launched and controlled from a regular PC computer that runs Windows (or sometimes Linux).

This is what helps light up the area around the plate to get the clearest view of the license. With any good-quality photo, it’s important to have lighting or flash lighting available to make the scene as bright as possible.

The computer running the application for the LPR system uses very specific software to carry out its functions. The most common software is a Dynamic Link Library, which recognizes the software and allows for full use.

The hardware carries out the software’s functions in the real world. It has input and output functions so that it has the best chances of performing flawlessly.

The frame grabber is a part of the program’s software and acts as a wall between the camera and the computer. It’s what allows the software to see the image. The frame grabber will convert the raw image data into an actual image for the user to see.

All images are immediately submitted to a database of records. They can be saved locally or sent via a network to a server. Usually the database contains images of the cars and sometimes the driver’s faces.

Typical LPR Systems

The illustration below highlights typical LPR system configurations. This specific one is labeled “SeeLane”, which is an application running in Windows. SeeLane provides the interface for the user, so that the user may control the SeeCarHead camera and illumination components of the LPR system. The frame grabber provides the interface for each unit, which is assigned one per vehicle. The application provides all the components that the user needs to control the sensors, lighting, camera, and more.

This advanced application can also show automatic results and simultaneously transmit these results over DDE messages to other applications it’s connected to. This makes it easier to track traveling vehicles over long distances or periods of time. Plus, you can determine which database all of this information is sent to.