Gini index based search space selection in Compressive Sampling Matching Pursuit

Compressive Sampling Matching Pursuit (CoSaMP) is a popular sparse recovery algorithm in Compressed Sensing. For a K-sparse signal, CoSaMP always selects a fixed number of atoms, 2K, from the matched filter in every iteration. Empirically we observed that this strategy is not efficient and deteriorates the performance of CoSaMP. To alleviate this drawback we propose to use Gini Index of the sparse signal as a measure to select the potential atoms from the matched filter. Using Monte Carlo simulations we show that the proposed modification improves the sparse signal recovery performance of CoSaMP.