Physicists Take Low-Light Images Using Less Than One Photon per Pixel | Hacker News Reader

Physicists Take Low-Light Images Using Less Than One Photon per Pixel

medium.com

75 pointsmarkfenton12 comments

You don't even need to capture a photon coming from the object.

Shoot a red laser at a couple of non-linear crystals to split the photons into a red/infrared entangled pair. Now, send the infrared ones through the object and into the other crystal while you divert the red ones to a screen without touching the object. As you can no longer determine which infrared photon belongs to what entangled pair the information they ones had is now contained in the red photons instead - and an image of the object appears on the screen.

http://medienportal.univie.ac.at/presse/aktuelle-pressemeldu...

Who the hell needs magic when there's quantum physics.

That's not quite what happens. Based on your description information about an object could travel faster than light.

You send light from cristal A to the object and then on to cristal B. You then construct the image using light from both cristals. The specific spookiness is only apparent when you try and track an individual photon.

This is similar to nuclear gamma-ray imaging, such as in medical application, where the number of counts is inherently low because the high energy photons are more penetrating and thus escape the camera.

Heralded imaging is explained here: https://en.wikipedia.org/wiki/Ghost_imaging

I wonder if this could be used to boost resolution of SEM and similar devices with electrons instead of photons. It'd make imaging non-metallic objects a lot easier since you won't have to bombard it with as many electrons, potentially destroying the object you want to examine.

Unlikely. Resolution limits are determined by aperture and wavelength, which is unrelated to photon number. There may be other tricks that can get around those limits, but they will be quite different from this sort of photon-number stuff.

Photon number is conjugate to phase, whereas to improve resolution you need to know phase very precisely, so the uncertainty in (and therefore the number of) photons goes up.

"Photon number is conjugate to phase' - I'd like to understand that better, is it the same as saying photon counting destroys phase information? Would it also hold true for gated counting?

Maybe it's just my eyeballs, but it looks like the <1 photon/pixel result was kind of fuzzy vs. the higher photon count image (cf. wasp wing).

If the idea is getting targeted bits of information about an object, as in whether a feature is present or absent, the very low photon image might meet the need.

However, with photography as art, the requirements for image resolution would be much higher, much more similar to the multi-photon image examples.

So it would appear mileage varies: the minimal photon images suit the minimalist domain, but an ordinary high-resolution image would require many more photons/pixel to get the expected result.

The paper on "heralded imaging" seems to be only available as a paywalled abstract, so I have no idea what that is. :(

Without seeing the visibility curve, these results could be purely classical.

Could this be used for telescopes?

No. This requires you to be in full control of the photons illuminating the image so you can properly entangle and detect them.

Reply on news.ycombinator.com