Scrambling light may lead to sharper images, wider views
"It allows you to take a closer look at an object without narrowing your field of view," said Jason Fleischer, an assistant professor of electrical engineering at Princeton who led the research. The study, co-written with graduate students Christopher Barsi and Wenjie Wan, is reported as the cover story in the April edition of Nature Photonics.
Cameras and other optical devices -- including the human eye -- are limited by the amount of light that they can collect through their lens openings, or apertures. In order for a light ray to be recorded, it has to pass through the lens and reach the device's "detector" -- such as the eye's retina or a digital camera's detector. But many light rays never make it to the detector, either because they are too weak, or because they are deflected.
This problem is particularly acute with details
that are smaller than the wavelength of light. (Each color of light has
a distinct wavelength -- green, for instance, has a wavelength of 530
nanometers, roughly the size of a typical bacterium's internal
structure.) Light rays from such tiny features fade before they reach
the lens. To capture these rays, devices have to probe very near the
surface of the object, and scan it point-by-point, stitching together a
full image.
"In effect, these devices suffer from 'tunnel vision,'" said Fleischer.
The new method addresses the shortcomings of small apertures by taking
advantage of the unusual properties of substances called nonlinear
optical materials. In conventional lens materials such as glass or
plastic, rays of light pass through without interacting with one
another. In nonlinear materials, light rays mix with each other in
complex ways. Rays that don't reach the camera may pass along some of
their information to rays that do get recorded by it. Thanks to the
mixing of rays, information that would otherwise be lost manages to
reach the camera.
The image from a nonlinear lens would therefore be rich in detail.
Unfortunately, it would also be distorted -- and useless for
conventional optics. But if the information could be unscrambled, a
computer could reconstruct a high-resolution undistorted image of the
entire scene. "In such an image all parts of the scene will be 'zoomed
in' at the same time," said Fleischer.
Until now, scientists have achieved this unscrambling and
reconstruction only in highly constrained settings such as fiber-optic
cables. This is partly because cameras and other optical equipment
typically don't record full visual information -- they record only
color and brightness, the properties that are perceived in everyday
experience. Much of the information essential to recovering the finer
details of an object or scene is captured by another light property
called phase, which measures the time and location of a wave peak.
Armed with these techniques, the Princeton team set up its imaging system. The core component, a nonlinear wave mixer, is a rectangular pill-sized crystal of a material called strontium barium niobate. The researchers placed the object to be imaged on one side of the crystal and image-capturing equipment on the other. They tested the system by obtaining images of various objects, including a chart developed by the Air Force that is widely used to calibrate optical devices. In each case the system could image the objects with high resolution. “We were surprised at how well it worked,” said Barsi.
By capturing information that would normally be lost, the new method could greatly enhance the resolution using normal light -- allowing scientists to build microscopes and other devices capable of so-called super-resolution. Conventional methods could attain similar resolution using ultraviolet light or X-rays, which have much smaller wavelengths than visible light, but this type of radiation harms living cells. "The new method could help you see with much better precision, but with a light the cell actually likes," said Fleischer.
Another interesting use could be in tomography, a technology often used to get 3-D images of body parts for medical diagnostics. Current methods typically take a number of 2-D images from different viewpoints and assemble them into a 3-D image. In contrast, the new system makes it possible to directly compute the 3-D visual information from images at a single viewpoint, potentially simplifying the setup.
Other potential uses of the new method are in data encryption and in characterizing the optical properties of new nonlinear materials. The method could improve scientists' fundamental understanding of how light behaves when passing through such media. In addition to exploring some of these possibilities, the researchers are working on developing a better nonlinear lens material and improving their reconstruction technique.
This research was supported by the National Science Foundation, the Department of Energy and the Air Force Office of Scientific Research. Barsi was supported by a National Defense Science and Engineering Graduate Fellowship provided by the Army Research Office.
Share / Save
