Stagewise Arithmetic Orthogonal Matching Pursuit

In order to improve the problems that stagewise weak orthogonal matching pursuit (SWOMP) has low reconstruction accuracy and imprecise choice of indexs selecting, an effective algorithm called stagewise arithmetic orthogonal matching pursuit (SAOMP) was proposed. As an extension of SWOMP algorithm, SAOMP algorithm first adopts an arithmetic threshold strategy to improve the accuracy of the selected indexs, and then introduces a backtracking step to flexibly remove some indexs wrongly selected at previous processing. Through these two modifications, SAOMP algorithm achieves superior reconstruction performance with low complexity. Simulation results demonstrate that, under the same condition, SAOMP algorithm can get better reconstruction result and higher exact reconstruction probability. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.