Hacker News new | past | comments | ask | show | jobs | submit login
‘Invisible barcodes’ could improve sorting of recycling [video] (bbc.com)
25 points by sohkamyung 31 days ago | hide | past | web | favorite | 23 comments

I've always wished that you could get a separate IR channel in phone cameras. Combine this with IR ink or IR channels on video feeds and you've got invisible QR codes. I suspect there's huge potential for being able to point your phone at content and have it magically link to data about the product / show / ad / event / whatever.

Phone cameras can see in IR. A classic trick to test if your TV remote is working is to look at the LED through a digital camera while pressing a button. Try it right now if you don't believe me.

What's missing is screens which can display meta information in the IR channel. I say we call the image format IRGB.

Yeah, I use it to see if my old remotes' batteries are dead.

They can see in IR, but they can't distinguish it from other light. Maybe that's good enough.

iPhones have IR filters over them so that doesn't work anymore unfortunately.

On my iPhone 7 at least, the rear camera seems to have an IR filter but the front one does not (or perhaps it has a filter but is less effective). So I can still test remotes in 'selfie' mode.

Doesn’t FaceID use structured IR light? If so, at least one of the cameras won’t have a filter.

The front camera usually doesn't filter IR.

They might be sensitive to IR without sufficient filtering but they're not selective to it, I think the idea here is something like a IR,R,G,B pixel format. Maybe you could get military interest in it (digital Aerochrome).

Also a good way to detect hidden surveillance cameras at night.

More generally, this problem is called Optical Camera Communication (OCC).

One of the more interesting things you can do is to exploit the rolling shutter readout method of consumer CMOS sensors to recover information with a high temporal resolution. By modulating the backlight of a video display or any other kind of wide-area light, you can transmit information optically to the phone in a way that's invisible to the user.

I'm not aware of any commercial applications of such tech, but there are enough keywords in this post to find a few research papers on the topic.

Great, now Facebook has yet another channel of data to mine your pictures for…

Isn't this a solved problem? I thought different plastics can be accurately differentiated by multispectral imaging?

i.e. that the problem with recycling isn't in identifying the plastic, it's in separating and cleaning it?

Plus recycling plastic is pointless - it's environmentally better to burn plastic for energy.

A big part of the problem is separating food vs non-food containers made with the same plastic - this tech can help with that by identifying the use even after the label has been removed and the package partially crushed.

> Plus recycling plastic is pointless - it's environmentally better to burn plastic for energy.

Source please? This seems very counter intuitive.


"For some products, recycling is an effective solution. Aluminium cans and glass, for example, are infinitely recyclable and can be reprocessed in the UK. But for plastics, it’s a different story.

For one thing, there are so many types of plastic it makes it hard to sort correctly. Consumers inadvertently mix recyclable and non-recyclable plastics in the same box, which contaminates the load and requires there to be further sorting and segregating, which not all collectors do, and effects the value and re-usability of the plastic when it’s resold. "


"Recycling plastic uses up a lot of resources, and after all the hauling around, sorting, and processing of bottles and containers, it often ends up getting thrown away or burned"

One 2017 report published in Science Advances suggested only 9% of the plastic that we ever use is recycled, while consulting firm McKinsey estimates just 16% of plastics are "reprocessed" and turned into new plastic goods.

People don't have much use for the small portion of plastic that does get recycled anyway. In the European Union, only 6% of plastic demand is for recycled plastics.

That didn't seem to actually try to support the statement.

Here's another UK based meta-study that suggest recycling is always better than incineration for plastic in the UK:


Those are 2010 numbers so I'd guess the lower carbon grid makes the case for recycling better.

In general incineration across multiple materials seems to do better than landfill but worse than recycling.

while glass is technically "infinitely recyclable", it almost never is recycled into glass. it requires far too much sorting and the cost the raw materials that go into glass (sand) is dirt cheap, so it is far less energy intensive to use fresh ingredients. Most glass gets crushed and turned into "cullet" and it is used as a gravel replacement for road beds.

It's too cheap when new, provides little value as a recycled product and is far easier than glass or metals to corrupt by getting contaminants in the batch. Contaminants could be as simple as a bottle of the wrong type of plastic or remnants of lid or labels.

Better to discourage plastic use through an escalating tax, use home compostable alternative plastics and burn what is unavoidably produced in the local combined heat and power (that most places don't have). Of course it's still a hydrocarbon so ideally you want carbon capture and storage with that.

The fundamental challenge here is creating a market.

If governments pay for recycling then they have the incentive to change it, but no easy levers to pull to make those changes.

If they instead just bill the producers of the packaging for the cost of recycling or landfill then there's an instant incentive for the people making the packaging to reduce, reuse and recycle more.

Then multiple ingenious ideas like this can fight it out in the market.

Very similar to my take on carbon taxes. We have the tech, but not the market to enable it to be deployed and developed further.

The project is called "Holy Grail", and the plastic embossing with the 'invisible barcodes' seems to be the cleverest part. Does anyone have experience with this sort of technology (using or developing)?

This is the company providing watermarking technology: https://en.wikipedia.org/wiki/Digimarc

What I came with: take 8 fluorescent dies with machine distinguishable fluorescence wavelengths, then you use them to code plastics. 2^8=256 enough for all common materials

There's no point in recycling most things curbside. Even in the places where the garbagemen don't grab the recycling bins and toss them in with the landfill trash, the things that curbside recycling takes are easier, cheaper, and better to just make new.

It's not like it's steel, or lead-acid batteries.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact