Pages

Sunday, June 08, 2008

CS: Another Single Pixel Camera at Rice.


Marco Duarte of the Rice Compressive Sensing Group stated in the comment section of the entry on the U. of Arizona FSS Imager the following:

We have also developed an active illumination-based single pixel camera using an off-the-shelf overhead projector and a laptop with a USB-based A/D device connected to a sensor. Our goal to developing this camera was portability of the system. In fact, we have presented demonstrations using this portable camera at the TI Developer's Conference 08 (Dallas, TX) and at ICASSP 08 (Las Vegas, NV).
[Marco kindly forwarded the photo of the set-up above with the mention that they "have not published material on the new camera since it is essentially the same architecture as the old one. The only thing that has changed is the active vs passive illumination..."]. The way I am looking at it, this would be the second single pixel camera at Rice but more importantly, since it looks simple and cheap enough it would make a great Hack A Day or Make entry. If you look at the Hack A Day statistics,
it would be a great way to get hobbyists to learn about Compressed Sensing as building the hardware is not the biggest issue.

4 comments:

  1. Hi all, I'm trying to implement a setup similar to the one you present here but I'm having some problems in replicating your results.

    Regarding the photodetection setup, do you have any constraints there? Or is it a simple photodetector with an amplifier?

    Any hints...? Help?

    Thank you very much in advance!

    ReplyDelete
  2. Hello Anonymous,

    Let me ask some of the people involved in the set-up.

    Cheers,

    Igor.

    ReplyDelete
  3. check out the work on dual photography from stanford which did essentially the same thing in 2005 (published in SIGGRAPH 2005). they provide more details on the implementation there.

    the authors also extended their ideas to compressed sensing as well.

    ReplyDelete
  4. As the anonymous commenter said, the difference between the "normal" set up and the CS one is really the pattern being used and the reconstruction. The hardware should not be different.

    Igor.

    ReplyDelete