Physicists Take Low-Light Images Using Less Than One Photon Per Pixel
How many photons does it take to form an image? Not very many, say the researchers revolutionising photography
The science of imaging is currently in the midst of a huge revolution. Researchers have recently discovered a number of innovative techniques for producing pictures in circumstances that would have seemed impossible just a few years ago. They’ve worked out how to take high-resolution pictures using a single pixel, how to make lensless cameras and how to photograph objects hidden behind other things.
These new techniques are pushing the science of imaging to its limits. But where do the limits lie?
Today, Peter Morris at the University of Glasgow in Scotland and a few pals ask a fundamental question: how many photons does it take to form an image? The conventional answer is tens of thousands of photons per pixel, at least in an ordinary camera.
So their breakthrough is something of a surprise. Morris and co have created images using less than one photon per pixel. That’s not just lowlight imaging, it’s almost pitch black. And it could have applications ranging from biological imaging, where photons can damage a sample, to covert imaging, where there is little light available to take photographs.
Their trick is to combine two recently discovered imaging techniques. The first relies on entangled pairs of photons and is known as heralded imaging. The idea here is to create a pair of photons and use one of them, the herald, to trigger a detector that records the other photon making an image.
The advantage of this technique is that the photon detector only operates at the instant the imaging photon arrives and that means background noise can be more or less eradicated. “The use of heralded single-photons ensures that the background counts can be virtually eliminated from the recorded images,” say Morris and co.
The second technique is known as compressed sensing. The idea here is that many ordinary measurements have huge redundancy. So it is possible to get the same result using just a small fraction of carefully chosen data points.
The trick is in knowing which measurements to take and how to reassemble them to create an image. All that is made easier when the rmeasurements follow some well-known statistical distribution, as is the case in imaging. When the distribution is already known, just a small number of data points can determine its properties.
Combining these techniques has allowed Morris and pals to explore the limits of photography. They begin by using the technique to image a standard US Air Force resolution target using an array of 300 by 300 pixels. “We see that we are able to form a reconstructed image of the test target for <7000 photons, which corresponds to less than 0.2 photons per image pixel,” they say.
They go on to use the same technique to make an image of a wasp wing. “The low photon number image comprises only 40419 detected photons over a field of view of 90,000 image pixels, corresponding to 0.45 photons per pixel,” say Morris and co.
These are impressive results — images taken using significantly fewer than one photon per pixel. And they could have a significant impact on the way images are made of light-sensitive subjects. In particular, biologists are well aware that biological samples can sometimes be significantly damaged by photons so the fewer used, the better.
All this provides an answer of sorts to the question originally posed: how many photons does it take to form an image? The answer is clearly less than one per pixel. But how far can physicists take this technique?
Clearly, there are more exciting results to come.
Ref: arxiv.org/abs/1408.6381 : Imaging With A Small Number Of Photons
Follow the Physics arXiv Blog on Twitter at @arxivblog, on Facebook and by hitting the Follow button below
