Jie Yan and Wu-Sheng Lu
Linearized Bregman, FISTA, Compressive Sensing
A central problem in compressive sensing is the recovery of a sparse signal using a relatively small number of linear measurements. The basis pursuit (BP) has been a successful formulation for this signal reconstruction problem. Among other things, linearized Bregman (LB) methods proposed recently are found effective to solve BP. In this paper, we present a fast linearized Bregman algorithm applied to a dual formulation that accelerates the conventional LB iterations considerably. Performance of the proposed algorithm is evaluated and compared with the conventional LB algorithm in compressive sampling of 1-D sparse signals and digital images.
Important Links:
Go Back
