A tree-based regularized orthogonal matching pursuit algorithm

Abstract

Reconstruction algorithm is a significant research field of compressed sensing (CS). Among existing algorithms, regularized orthogonal matching pursuit (ROMP) enjoys the merit of implementing fast recovery procedures. Recent studies have recognized that sparse signals have special sparse structure, which is useful for reconstruction as prior information. In this paper, by utilizing the sparse tree structure as prior information, we propose a tree-based regularized orthogonal matching pursuit (T-ROMP) reconstruction algorithm. Furthermore, we set a ratio factor to reduce the error probability of the support set. Compared to ROMP, simulation results indicate that the proposed algorithm achieve better reconstruction performance for different conditions.