Friday, December 30, 2011

It's not a spinning disk either ...

Remember when a year and a half ago, I mentioned that the 'thing' that enables one to render compressive measurement universal and timeless should not be called a light modulating device or a rotating collection of masks, but rather a Randomized Carousel.

Mad Men ´The Carousel´ from Emilio on Vimeo.

Well, it looks like we have an instance of one: Spinning disk for compressive imaging by H. Shen, L. Gan, N. Newman, Y. Dong, C. Li, Y. Huang, and Y. C. Shen. The abstract reads:

We report the first, to the best of our knowledge, experimental implementation of a spinning-disk configuration for high-speed compressive image acquisition. A single rotating mask (i.e., the spinning disk) with random binary patterns was utilized to spatially modulate a collimated terahertz (THz) or IR beam. After propagating through the sample, the THz or IR beam was measured using a single detector, and THz and IR images were subsequently reconstructed using compressive sensing. We demonstrate that a 32-by-32 pixel image could be obtained from 160 to 240 measurements in both the IR and THz ranges. This spinning-disk configuration allows the use of an electric motor to rotate the spinning disk, thus enabling the experiment to be performed automatically and continuously. This, together with its compact design and computational efficiency, makes it promising for real-time imaging applications

While the paper is behind a paywall, here are the videos showing some results of reconstruction using:

TV-min nonlinear algorithm
 MMSE linear estimation

This will be added to compressive sensing hardware shortly.

Credit: AMC, Mad Men.
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.


Anonymous said...

BTW, I recently learned that the spinning disk is used in confocal microscopy for making the laser sample scan faster [1]. No randomness there... for the moment.

The idea traces back to 1920 with the Nipkow disk [2]



Igor said...

Good catch Anonymnous.


Mike Hughes said...

It looks like this could be a solution to getting acceptable frame rates from CS cameras.

What I find most interesting about this is that the correlation between adjacent frames doesn't appear to affect the image quality very much - there are no obvious artefacts.

I guess the drawback of this approach is that it is harder to know exactly what mask you are applying at any one time. With an SLM you know what mask you have applied as it's whatever you have sent to the SLM.