Tuesday, February 02, 2016

Thesis: Electronics Design and Implementation of a Compressed Sensing Instrument for Astronomy, Raquel Bandarra Borges

If anyone knows Raquel, I'd love talking to her !

Raquel uses a TI lightcrafter for the light modulation and an arduino to do the synchronization with the single pixel. She uses this Matlab to Lightcrafter library written by Jan Winter at TU Berlin (on top of TI's instruction sets). There is a related set of Python library by Evan Moore at Google.

Electronics Design and Implementation of a Compressed Sensing Instrument for Astronomy by Raquel Bandarra Borges

Compressed sensing is a new concept that is attracting much research nowadays. The theory states that under certain conditions, only a relatively small number of samples need to be acquired for reconstructing an image or any signal in general [1]. In other words, compressed sensing enables signal compression at the acquisition stage. In contrast, with traditional methods the data is compressed after acquisition. An outstanding characteristic of compressed sensing (CS) is that signals can be reconstructed at sampling rates lower than those required in the framework of the Shannon-Nyquist sampling theory [2, 3]. This theory has been increasingly developed, in areas like computational mathematics, signal processing, among others [4]. Given that modern sensors are producing more and more information, applications of CS start to appear in areas where data compression is an essential requirement. The most known early application is the Rice University single pixel camera [5]. Now, there are other instruments based on CS and applications to communications, medicine, astronomy and others research fields.
In this work we designed and implemented the electronic acquisition and control subsystems of a compressive imaging single pixel camera. The subsystem developed is composed of a Digital Micromirror Device (DMD) for coding the signal, a controller subsystem, an acquisition subsystem and a storage subsystem. These last two subsystems are integrated in a printed circuit board. The most important features of the system that we designed in this work is tahta it can provide good resolution with fast acquisition times, that it ensures synchronization between the modification of the signal encoding patterns and the signal acquisition, and that it is relatively a↵ordable and easy to reproductible. The DMD is used for creating pseudorandom patterns that allows to encode an image with fewer samples than the image’s number of pixels. This camera uses only a single detection element, enabling the reduction of the complexity and the cost of the system. Similar systems can be used for high resolution imaging in wavelengths for which currently instruments are expensive, bulky or even impossible to build.

Join the CompressiveSensing subreddit or the Google+ Community or the Facebook page and post there !
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.

No comments: