Skip to Content

New Camera Can See Around Corners

Chinese scientists have built a camera capable of photographing objects it can’t directly see.

In the last few years, single-pixel cameras have begun to revolutionize the field of imaging. These counterintuitive devices produce high-resolution images using a single pixel to detect light. They do not need lenses, the images of have none of the distortions that lenses produce, and the entire picture is always in focus. Physicists have used them to make movies and even to create 3-D images.

And that raises an interesting question: how much more can these devices do?

Today we get an answer of sorts thanks to the work of Bin Bai and co at Xi'an Jiaotong University in China, who have built a single pixel camera that can see around corners. Their new device can photograph objects even when they are not in direct view.

The technique is similar to that used with other single pixel cameras. The trick is to first randomize the light that the pixel detects, record the resulting light intensity, and then repeat this process thousands of times.

It’s easy to think that this randomization makes the task of creating an image even harder, but the reverse is true.

The randomization process changes the intensity of light each time the pixel records it. These differences in intensity are not random but instead correlated with the scene in front of the pixel. So producing an image is simply a question of mining this data to find the correlation. And the more data that is collected, the better the image becomes.

So by recording the intensity of light many times, it is possible to create a high-resolution picture with a single pixel.

And that’s how Bin and co have used their camera to peer round corners. What they mean by this is that they have produced a picture using light from an object that is scattered off a wall. The pixel records the scattered light but cannot see the object directly. 

The setup is simple. The team begins by illuminating a toy airplane with light from a beamer that projects a random pattern of illuminated squares. This random pattern changes with each exposure.

The airplane is next to a white diffuse wall that scatters light toward a single pixel that cannot otherwise “see” the plane. This pixel records the light intensity from the wall. The team repeats this imaging process some 50,000 times. Finally, the team uses a data-mining algorithm to crunch the resulting data set and to create the image of the plane shown here.

The results are impressive. “All results show that the single-pixel camera can image the diffuse object when the camera cannot look at object directly,” say Bin and co.

The airplane is clearly visible in the image and the team says further improvements are possible by, for example, optimizing the algorithm and by reducing the size of the squares in the random pattern of projected light. “When the size of speckle patterns becomes smaller, the resolution of the image becomes higher,” say Bin and co.

An interesting question is how quickly it is possible to take 50,000 images, since this places an important limit on how the camera can image moving objects.

The Chinese team doesn’t suggest potential applications for the device, but it’s not hard to think of ways it might be used. Single-pixel cameras are tiny, simple to make, and cheap to build.

One way or another, single-pixel cameras are set to have significant impact in the not too distant future.

Ref: arxiv.org/abs/1612.07120: Imaging Around Corners with Single-Pixel Detector by Computational Ghost Imaging

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.