Unlike regular digital cameras that capture multiple pixels simultaneously, this new technique involves imaging a single pixel multiple times.
Now here's an interesting idea. Let's start by considering a conventional one megapixel digital camera. In this case, the main sensor can capture a million pixels of data simultaneously. The data associated with the million pixels is then heavily processed to compress it into a manageable form.
Well, some cunning little scamps at Rice University have come up with an alternative technique, based on a Single Pixel Sensor. First of all they have a silicon chip covered with hundreds of thousands of microscopic mirrors, each the size of a single bacterium.
These mirrors are used to focus light onto a one-pixel sensor. In effect, they capture several of thousands of versions of this single pixel in rapid succession. What? Did I just say "several thousand"? How does this equate to a megapixel? Well, this is where things start to get really clever.
The trick is that the mirrors are shuffled at random for each new sample. In effect, they are performing their version of a compression algorithm, but in a radically different form to a conventional camera. This is the reason the litttle rascals at Rice call their technique "compressive sensing".
So why is this of interest? Well, lots of reasons actually. But just to give you one example, when you start imaging things outside of the visible spectrum, it can be expensive to create large arrays of detectors. Using the Rice approach, you need only construct a single sensor. Pretty cool eh?
For more information and photos, visit: http://dsp.rice.edu/cs/cscamera.
Questions? Comments? Feel free to email me – Clive "Max" Maxfield – at email@example.com). And, of course, if you haven't already done so, don't forget to Sign Up for our weekly Programmable Logic DesignLine Newsletter.