WHY INFRARED FOR ANPR??

Read… be informed… understand!

Think about it

And you can always contact us and ask questions

So, why do they use infrared wavelengths for ANPR? Go grab a coffee, you’ll be here a while….

Yes, sorry, but this is where we get technical. You want to understand all this traffic imaging and ANPR then you have to get technical. And there is a lot of technical stuff, I’m afraid. None of this is simple and much of it depends on pretty complicated science. How much physics do you remember from school?!

There will be one or two generalisations here, but read it all, ponder it, and you will have a much better idea of why the cameras use infrared light to take pictures of your number plate.

First place to start is to understand what infrared wavelengths are and how they can be detected.

Light is a wave and it is firmly characterised by something called wavelength. Light wavelengths are measured in nanometres or nm. The sun constantly emits a range of wavelengths from about 250nm to 3500nm, with its maximum energy output at about 550nm. Something like this:

Wavelengths of light emitted by the sun

The sharp dips in this graph are places where light is strongly absorbed by the earth’s atmosphere. In outer space these dips do not exist but because of the atmosphere, on the earth’s surface these wavelengths do not reach us.

The human eye has evolved to respond in the region of the sun’s output where most light is. That is from about 400nm to 700nm. This is the “visible” or “colour” spectrum. 400 to 500 is blue, 500 to 600 is green and 600 to 700 is red. Below 400 is the ultraviolet, above 700 is the infrared. The sensitivity of the eye varies with how dark it is but basically the peak response is in the green region:

The Sensitivity of the Human Eye

Big thing to notice here is that our eyes do not respond to infrared wavelengths – we cannot see in the infrared. However, a silicon chip camera can….

The massive preponderance of cameras these days is down to the huge expansion in the manufacturing capability and capacity for the digital imaging chip, more properly known as the ‘charge coupled device’. This was driven by the mobile camera phone market. These camera chips are now made in such huge volumes that they appear in all sorts of places. I mean, 25 years ago we’d have laughed at the idea of a doorbell camera! These cameras are now so cheap that the manufacturers are going chasing markets for them.

The silicon chip camera does not work like your eye. The chip has ‘pixels’ defined into the silicon and these pixels absorb light. The electrical signal generated when the light is absorbed is drawn out and digitally recorded. The result is a digital image file.

These pixels respond to a range of light wavelengths which is far greater than the eye:

The Response of two Types of Silicon Chip Camera

Take a close look at this graph. It shows that a silicon chip camera can ‘see’ from down at about 300nm up to about 1100nm. Much more than your eye can see.

In fact, a chip camera can see so much light that if you want it to produce a colour image that mimics the view of the human eye, you have to block the other wavelengths. You have to put a wavelength blocking filter over the chip and block some regions of light wavelength. Don’t forget that a human eye can only see blue, green and red, or 400nm to 700nm…

If they don’t do that, then when a digital image is created, all the brightness and contrast is messed up. If you want a camera to produce an image which looks real to our eyes, you must block the wavelengths that we can’t see.

A chip camera which produces a colour image has had the infrared (and ultraviolet) wavelengths hitting the chip removed.

But that goes both ways…

If you block 400nm to 700nm, the chip can ‘see’ in the infrared region of the spectrum. And it is this fact that makes it VERY useful for ANPR imaging.

OK, so we have discussed how the human eye works, how a chip camera works and that chip cameras can be made to ‘see’ wavelengths of light that we can’t see. How does that help with automatic number plate recognition?

ANPR is a computer methodology, an algorithm or basically just a computer programme. Its purpose is for extracting letters and numbers from digital images. 

The powers that be want to read your number plate, right? But they don’t want to employ thousands of people watching screens all day. Costs too much. They want to do it with computing power. In order to do it with computing power they need a digital image which has the very best resolution. They need an image of your number plate which is as clear and well defined as it is possible to get. Whether they ‘read’ your number plate or not depends on how good the image is that they apply the ANPR algorithm to.

If a chip camera is made to generate a colour image which is like that of our eye then it does NOT generate an image which is the clearest for applying an ANPR algorithm.

This is where the physics gets complicated. We can’t go too far into this (I’ll be surprised if most people get this far with all this technical shit!), but here’s a list of the reasons an infrared image gives a better picture for a computer to read from:

So, there are the advantages of infrared ANPR imaging and there’s a whole industry grown up around it. You want an IR floodlight for your ANPR camera? Try these:

These ultrabright LEDs emit infrared wavelengths from about 830nm to 870nm. Way above what the eye can see and just in the right place to gain all the advantages mentioned above.

You want a filter to cut our all wavelengths except the floodlight? Try these:

Put this filter in front of your chip camera, illuminate the road with a floodlight made from Ushio LEDs, then you have an infrared ANPR camera. And that camera produces very high resolution images of your number plate that look like this:

That’s how stark an infrared ANPR image is. Little resolution beyond the number plate, which is very clear and bright – the ANPR computer algorithm has the best chance of capturing this registration