[For those of you who do not know about compressive sensing, it is the technique that is behind, among other implementations, the single pixel camera or the "cheap" hyperspectral imager (CASSI). One of the main challenges of this technology is the slow reconstruction process whereby an image is created out of the measurements made by the camera.]
Alexandre Borghi, Jerome Darbon, Sylvain Peyronnet, Tony F. Chan and Stanley Osher just released a paper on the possibility of reconstructing a signal from compressive sensing measurements using multi-CPUs, GPUs and the Cell processor in a paper entitled: A Simple Compressive Sensing Algorithm for Parallel Many-Core Architectures, The abstract reads:
In this paper we consider the l1-compressive sensing problem. We propose an algorithm specifically designed to take advantage of shared memory, vectorized, parallel and many-core microprocessors such as the Cell processor, new generation Graphics Processing Units (GPUs) and standard vectorized multi-core processors (e.g. quad core CPUs). Besides its implementation is easy. We also give evidence of the efficiency of our approach and compare the algorithm on the three platforms, thus exhibiting pros and cons for each of them.
It is an intriguing paper as most papers trying to deal with GPUs aim at implementing very generic solvers. In this case, the algorithm of the solver is taylored to the hardware available. The figure shows the time it takes for reconstruction for a signal where m/n = 12.5% . The number of non-zero element is m/10. The measurement matrix is a partial DCT.
[ To find out more about Compressive Sensing, please check either the Rice Reference site or Compressive Sensing: The Big Picture ]
No comments:
Post a Comment