Noise mitigated compressed sensing

Recently, compressed sensing (CS) is of major interest in the area of communication and measurement. CS technique is a subtle mathematical application in practice, which facilitates the signal acquisition and signal processing dramatically. It consists of the two phases: signal projection and signal recovery. Regarding the signal recovery often it is an l₁ optimization process in terms of a sparse regularized least squares. In this work, we introduce the noise-mitigated least squares (NMLS) to improve the CS signal recovery performance in case of suboptimal regularization parameter λ. Both theory and empirical results show that NMLS is a promising method over state-of the-art standard regularized CS procedures.