A standard digital camera uses a large number of photo sensors to determine the amount of light in each area of the image, referred to as a pixel. For example, a 10 megapixel camera measures 10,000,000 pixels using a CCD or CMOS light sensor. This works very well for optical light and is relatively cheap because of the maturity of CCD and CMOS technology. However, since CCDs and CMOS imagers only work in the optical range, infrared or ultraviolet cameras can be much more difficult and expensive to produce.
One solution to this is to use a single pixel camera. A single pixel camera uses only one light sensor to measure the entire image. This allows the use of one really good light sensor as opposed to 10 million very cheap ones. Compressed Sensing is used to measure the entire image using only a single sensor. Compressed Sensing is a mathematical process by which a signal can be undersampled (below the Nyquist rate), and recovered by using ℓ1 minimization, given that the original signal can be represented as a sparse signal and that the sampling process is incoherent to the sparse signal.
A single pixel camera works by reflecting the image off of a digital micromirror device (DMD) through a lens and onto a photodiode. A DMD is an array of microscopic mirrors which can be individually tilted ±15°. The micromirrors are tilted in such a way as to allow only some of the pixels to focus on the photodiode at any given time, while the rest are focused onto a light absorber. By doing this thousands of times and measuring the intensity of the light for each different mirror arrangement, we can physically realize the sampling matrix of a compressed sensing system. Any radiation which can be reflected from a mirror (including UV and IR) can be sampled this way, allowing for a UV or IR camera much cheaper than is currently available.
The main drawback to this system is that the measurements must be taken in series rather than all at once as in a traditional camera. Assume that, for a 1024 x 1024 image we need 25% of the pixels for a total of about 260,000 total pixels. The DMD can change positions in (at worst) 20 μs, allowing the entire image to be captured in 5.25s. The optical sensors stabilize much faster (around 20 ns for IR, 4 ns for optical), so there is potential for the imaging to become faster if the switching speed of the mirrors increases. However, even at this speed the camera could be used for many non-video applications.
* This page is informational only. The related projects are no longer under active development.