Tuesday, January 29, 2008

Compressed Sensing: Computational Optical Imaging and Spectroscopy videos, I need some more popcorn.

On the previous entry, we saw how Compressed Sensing (or Compressive Sampling) can be included in optical systems. In order to understand how the field has evolved to that level of integration of CS one may want to watch the videos of David Brady in 2005 in a workshop on Radar and Optical Imaging that took place at IMA in 2005 where he presented a week of talks on Computational Optical Imaging and Spectroscopy. The slides are here whereas the videos are here:

Some additional information can be found here. A very nice explanation on how a hologram is generated can be found at the beginning of presentation 9. Presentation 10 is also useful in order to better understand previous work at DISP. I liked the response to a question when he talks about a hierarchical set of lenses at 33 minutes into the talk 10 as it touches on a question asked in a previous entry. One also gets an insight as to why the one wants to use multiple apertures and why in focal plane coding CMOS technology might bring an advantage by allowing pixels to be distributed "randomly". I also note that there is no reference to NMF as regards to imposing non-negativity constraints as done by Bob Plemmons for some specific hyperspectral problems.

Credit photo: Wikipedia/National Nuclear Safety Administration, Operation Plumbob.

No comments: