Bokodes - amazing gizmos or a world-class hoax?

[kaled]

[news.bbc.co.uk...]

A replacement for the black and white stripes of the traditional barcode has been outlined by US researchers.

Bokodes, as they are known, can hold thousands of times more information than their striped cousins and can be read by a standard mobile phone camera.

The 3mm-diameter (0.1 inches), powered tags could be used to encode nutrition information on food packaging or create new devices for playing video games.

The work will be shown off at Siggraph, a conference in New Orleans next week.

The news media seem to think these things are for real, but in the absence of a credible explanation of the technology, I'm going to stick my neck out and say it's a massive hoax or some kind publicity stunt.

For example, they say, the tags are smaller, can be read from different angles and can be interrogated from far away by a standard mobile phone camera.

Given that these devices are only 3mm in diameter, that would represent a few blurry pixels - just how much information can a few pixels convey?

Kaled.

[oybama]

I too am skeptical, especially of the eventual cost of such a device. The current device is active (i.e. requires a battery) and emits a 3mm wide beam of light that contains some 2D coded information. The idea is that you defocus your camera (which can be difficult), which will cause a 3mm light source to grow to a much bigger spot on the camera imager. So you end up with a big blurry image instead of the small focused (or small blurry) image. Then you have some software in the camera that deblurs the image to yield a blown-up version of the 2D code. Eventually they hope that one could use the flash from a camera (or another external light source) instead of the current active LED. The flaw for things like cell-phone cameras is that they currently try to focus on objects, not defocus them. I wish them luck. We used cellphone cameras with bigger in-focus 2D tags and they work just fine.

[engine]

There are things such as the Microsoft Tag and QR Code [webmasterworld.com], but it's a pretty large code image to scan.

[kaled]

The idea is that you defocus your camera (which can be difficult), which will cause a 3mm light source to grow to a much bigger spot on the camera imager. So you end up with a big blurry image instead of the small focused (or small blurry) image.

It's twenty five years since I studied Fourier theory, but I can say with 99.9% certainty that will never work.

Unless the laws of physics and maths have changed recently, for anything remotely like the system described to work.
1) The information broadcast would have to be spectrum-limited.
2) It would have to be encoded over time not space.

Or, to put it another way, it would have to operate like a TV remote control, so that rules out many of the claimed features, including the ability of standard cameras to read them.

Maybe they could be solar-powered and maybe they could be triggered to broadcast by an anti-red-eye flash or something like that, but even with a software patch, no standard camera could ever read something like that.

Kaled.

[oybama]

MIT has some video describing the process and it shows (perhaps a simulation, perhaps real, perhaps smoke and mirrors) what they are doing.

They don't mention which algorithms they use to deblur the picture or what assumptions they are currently making that may or may not work out in real life. I would rate the presentation as propaganda, not engineering or science. You can find it here:

[web.media.mit.edu...]

[kaled]

Thanks for the link...

Well, it would seem I should eat a spoonful of humble pie, but maybe not a full portion. It looks like there is some genuine maths and physics here, however...

1) To discriminate a readable pattern, in practice a filter will be needed over the lens to remove the noise of ambient light (with the bokode light being spectrum-limited).
2) It appears the camera lens has to larger than those in typical cheap cameras and mobile phones.

Having skimmed the article, it would also appear that a decent quality lens may be required to discriminate between infinity and nearby, however, use of a filter may negate this.

If they move to infrared light (so that they can be bright without irritating people) and accept that standard cameras are not going to be up to the job of reading bokodes in practice, they may have something. For supplying information, they might be useful on shelves, etc but it's hard to see them being used on packaging.

Barcodes and RFID tags were both developed to fulfil a demand, whereas bokodes are a novel technology looking for an application.

Kaled.