Comparison of Generalized and Classical Reweighted Algorithms for Recovering Sparse SignalsCP6

We introduce a generalized l1 greedy algorithm for recovering sparse signals and demonstrate its superior performance over the classical reweighted l1 minimization algorithm (Candes, et. al) and the l1 greedy algorithm (Petukhov and Kozlov). Moreover, we show our algorithm is better at detecting small entries of unknown sparse signals thereby dramatically speeding up their recovery via l1 minimization. Finally, we discuss how to improve our algorithm by adapting the wavelet technique of semisoft thresholding.

This presentation is part of Contributed Presentation “CP6 - Contributed session 6

Fangjun Arroyo (Francis Marion University)
Edward Arroyo (School of Professional Studies, Northwestern University)
compressed sensing, image reconstruction, l1 greedy algorithm, l1-minimization, reweighted l1 minimization