Speaker: Fred Noo

Title: A brief introduction to compressive sampling theory

Reference:

Candes Publications
DSP

Abstract: Over the last 5 decades, most developments in signal and image processing have been based on Shannon sampling's theorem, assuming that the maximum accuracy that can be expected in the reconstruction of a signal is essentially dictated by the Nyquist frequency.

Recently, Emmanuel Candes from the California Institute of technology introduced a new sampling theory which he calls compressive sampling theory. This theory shows that signals can be very accurately reconstructed from much fewer samples than allowed by Shannon's theory, the gain depending on the class the signal belongs to.

Candes' theory looks at first glance wild and suspicious, however the mathematical community has now fully endorsed it and the signal processing community is clearly stepping in, so much that Candes' theory is now viewed by some as the most revolutionary theory in signal processing since Shannon's sampling theorem.