Thursday, May 10, 2007

Sparsify, another reason you should not avoid Compressed Sensing

Make that five codes (matlab) available to do sparse signal approximation. Thomas Blumensath just made available Sparsify:

Sparsify is a set of Matlab m-files implementing a range of different algorithms to calculate sparse signal approximations. Currently sparsify contains two main sets of algorithms, greedy methods (collected under the name of GreedLab) and hard thresholding algorithms (collected in HardLab)
Algorithms implemented include:
  • Orthogonal Matching Pursuit (in several flavors),
  • Matching Pursuit algorithm, Greedy Pursuit,
  • Approximate Conjugate Gradient Pursuit,
  • Nearly Orthogonal Matching Pursuit with partial Conjugate Gradients,
  • Iterative hard thresholding algorithm.
While reading his webpage, I noticed the INRIA's Matching Pursuit ToolKit project maintained by Sacha Krstulovic and Rémi Gribonval

Reference: Rice Compressed sensing site.

No comments:

Printfriendly