There has been a spark of interest over these nifty little machines recently. As the use of number plate readers increase over the next few years there are going to be more questions from the public about that they do and what they are for, not to mention more scaremongering from those who might have to believe this is one more steps towards a 1984 dystopia.

Did you know that number plate readers were actually invented in 1976, but have come a long way since then. When previously technology was limited and things like light, vehicle speed, how the plates were spaced and even what angle they where captured could skew the reading, now some clever clogs has perfected the system that eliminates most common inaccuracies.

Since then number plate readers, or Automatic number plate recognition (ANPR) to give them their proper name, have been impeccable public servants.

How number plate readers work …

Number plate recognition involves capturing a video or an image of a vehicle’s license plates and running them through a system of algorithms that converts the image to a text entry into a computer. The computers can then look up and all information in the vehicle’s history.

This is difficult to programme if you think about it. If you look at CAPTCHA images for example you can see why. Computers can only read data – as in text entry – and when confronted with an image it can only make out a bunch of pixels. Unlike the human eye it cannot pick out patterns and can therefore no read like we can. This is why this is so difficult and it takes a number of defining rules to make it possible.

1. Localisation

This algorithm determines which part of the vehicle the number plate reader looks at. Much like facial recognition of your camera phone, the localisation rule identifies key features of the vehicle and rules them out. For example, the bumper, the headlights, the mirrors, etc. Once these features have been identified and then ignored the camera is left with only the number plate to look at.

2. Sizing and Orientation

This part of the number plate reader’s algorithm accounts for distance and angular skews that may distort the image. This is the sort of thing you might see used alongside CCTV footage. It takes an image that is off angle and adjusts it into regular size and appearance. Obviously this correction makes the characters easier to read.

3. Normalisation

Like sizing and orientation this algorithm you might have previously seen alongside CCTV footage, however this corrects for blur, colour, brightness and contrast. Once again this makes the registration easier for the number plate reader to make sense of.

4. Segmentation

Faced with what we might think of a standard, front-and-centre picture of a number plate the ANPRs can really start getting to work. This part also makes it clear exactly why we have strict rules when it comes to how number plates can be displayed.

How segmentation works is be defining the boundaries in which the computer expects a character to appear in. It figuratively draws a box around each letter and uses those boxes to work out each letter individually. This is why the DVLA come down hard on anyone who changes the spacing or font on a number plate, since this stops the number plate reader doing its job!

5. OCR or Optical Character Recognition

This is the part where humans have to help a bit. Once the machine has isolated a character it is essentially left with a box of pixels that don’t make a lot of sense. Humans have to tell the number plate readers how to turn the pixels into letters by pre-programming the patterns to be expected. This is why font and size are important.

Upon recognition of these letters the computer can match the pattern to a specific letter, thus triggering an actual text entry. This obviously makes the searching and the reporting effortless on the technological side of things.

6. Analysis

This final step is so fast it is barely even a step. Using the number plate in text form it can check certain characters against their position on the number plate to check for age, area of registration and much more. It can use this to not only look up vehicle history but also add to that history. It literally takes milliseconds to do, but as you can see this function is the whole point of number plate readers in the first place.

We are now at a time when ANPRs are incredibly useful to society and really are not anything new to be scared of. In fact, they have been in wider-scale use since the 1990s and these modern versions have been very successful helping the DVLA and the police.

So why the bother? Probably because UK the most watched population in the world. In this country we have nearly 5 million CCTV cameras, which makes us the most watched population in the world. They say most Britons can’t go a regular working day within being filmed by 300 CCTV cameras. Feeling frightened? Feeling safe? No matter what side of the argument you fall on, CCTV and number plate readers are here to stay.

———-

By Sam Ryder. Sam is National Numbers’s resident busy-body and writer. She is new to the game but she is learning fast. Keep a look out for her other content, as well as her writing which you can find across the internet.

The post How do car number plate readers work? appeared first on National Numbers Blog.