Imaging by Compressive Sensing: A 1-Pixel Camera and Beyond
 |
|
Kevin Kelly
Assistant Professor of ECE
|
Compressive sensing is an emerging field based on the
revelation that a small number of random linear projections of a signal
or an image contain enough information for reconstruction of a high
resolution one. This technique is already being applied to traditional
imaging devices such as magnetic resonance imaging and neutron
scattering. Inspired by the success of this technology, we have
developed of a new optical imaging methodology that directly exploits
compressive sensing. The camera we have built is capable of megapixel
images while utilizing a single optical detector for acquisition. We
have directly implemented compressive sensing in the acquisition
through the use of a digital micromirror device to randomly modulate
and acquire the necessary projections of the image. Benefits of our
scheme include imaging beyond the visible spectrum where high
resolution image sensors arrays are much more costly as well as the
rapid acquisition of multispectra information. Our design has
additional desirable properties including scalable output bit stream
and variable image resolutions.
Thursday, September 13, 2007
3:00 p.m. - McMurtry Auditorium, Duncan Hall
2007 ECE Affiliates Conference
Global Partnerships in Research and Education