A single-pixel digital camera, scientists at Rice University
believe, will reduce power consumption and storage space without
sacrificing spatial resolution. The new approach aims to confront
one of the basic dilemmas of digital imaging, namely the huge waste
Consider that a megapixel camera will, when you take the
picture, capture and momentarily store a million numbers (the light
levels from the
No camera can store that much information for hundreds of
pictures, so an immediate data compression takes place right there
inside the camera.
A tiny microprocessor performs a Fourier
transform; that is, it converts the digital image into a weighted
sum of many sinusoid waves.
Instead of a million numbers, the
representation of the image can now been compressed into something
like 10,000 numbers, corresponding to the most important
coefficients from the mathematical transformation.
These are the
numbers actually retained for later processing into pictures.
The Rice camera saves space and energy by eliminating the first
step. It gets rid of the million pixels.
Instead it goes right to a
transformed version (about 10,000 numbers rather than a million) by
viewing the scene prismatically with a single pixel.
No, the light
from the object doesn’t go through a prism, but it is viewed about
10,000 different ways.
The light, in a quick succession of glances,
bounces off the myriad individually driven facets of a digital
micromirror device, or DMD
mirrors of a DMD (only a micron or so in size) do not image an
object or record data but merely steer light; they can be
individually angled in such a way that the light strikes a photo
detector or not, depending on whether the light is representing a
digital 1 or 0 at that moment.
The main idea is that the DMD is acting as a sort of analog optical
Each time the pixel views the object, a different set of
orientations is imposed on the array of micromirrors.
And, in an
interesting twist, the Rice camera uses random orientations.
Looking like the haphazard splotch of black and white squares of a
crossword puzzle, the DMD’s surface is reflective here and dark
there; some of the mirrors will faithfully reflect light from the
object to the pixel while others will, in effect, appear black.
Then the object is viewed again with a different micromirror
activation pattern; again the pixel will record an overall light
This process recurs about 10,000 times. Later, offline on a
computer, the single pixel light levels,
along with the micromirror patterns are processed using new
algorithms to reconstruct a sharp image.
This isn’t quite the old
type of imaging process, the kind used in x-ray crystallography or
CAT scans (which also convert pinpoints of data into images), but a
new kind of imaging called compressive sensing that is only about
two years old.
To summarize, the acquisition of imaging data is reduced many-fold
(saving on data storage), only a single pixel is needed (freeing up
valuable space in the primary detector), and the bulk of the
processing can be offloaded to a remote computer rather than a chip
inside the camera, thus greatly reducing power needs and extending
the usefulness of batteries.
Rice researchers Richard Baraniuk
Kevin Kelly (email@example.com) say that an additional virtue of the
camera is that with only a single pixel, the detector (a photodiode)
can be as fancy as you want.
It can even accommodate wavelengths
currently unavailable to digital photography, such as x ray,
terrahertz waves, even radar.
A working camera prototype has been
One of the main tasks is to reduce the time it takes to
record an image; the price for compressing space, pixels, and power
is to spread everything out in time since the cyclops-like pixel
must blink ten thousand or more times to capture the image.
Baraniuk says, the Rice form of photography is multiplexed in time.
The Rice results were reported last week at the Frontiers in Optics
Meeting of the Optical Society of America (OSA) held in Rochester
(www.osa.org/meetings/annual/) (For a picture of the setup and the
imaging results, see the web page http://dsp.rice.edu/cscamera and
the research paper at