Formulated as a least square problem under an ℓ 0 constraint, sparse signal restoration is a discrete optimization problem, known to be NP complete. Classical algorithms include, by increasing cost and efficiency, Matching Pursuit (MP), Orthogonal Matching Pursuit (OMP), Orthogonal Least Squares (OLS), stepwise regression algorithms and the exhaustive search. We revisit the Single Most Likely Replacement (SMLR) algorithm, developed in the mid-80's for Bernoulli-Gaussian signal restoration. We show that the formulation of sparse signal restoration as a limit case of Bernoulli-Gaussian signal restoration leads to an ℓ 0-penalized least square minimization problem, to which SMLR can be straightforwardly adapted. The resulting algorithm, called Single Best Replacement (SBR), can be interpreted as a forward-backward extension of OLS sharing similarities with stepwise regression algorithms. Some structural properties of SBR are put forward. A fast and stable implementation is proposed. The approach is illustrated on two inverse problems involving highly correlated dictionaries. We show that SBR is very competitive with popular sparse algorithms in terms of trade-off between accuracy and computation time.

Tropp's analysis of Orthogonal Matching Pursuit (OMP) using the Exact Recovery Condition (ERC) [1] is extended to a first exact recovery analysis of Orthogonal Least Squares (OLS). We show that when the ERC is met, OLS is guaranteed to exactly recover the unknown support in at most k iterations.Moreover, we provide a closer look at the analysis of both OMP and OLS when the ERC is not fulfilled.The existence of dictionaries for which some subsets are never recovered by OMP is proved. This phenomenon also appears with basis pursuit where support recovery depends on the sign patterns, but it does not occur for OLS. Finally, numerical experiments show that none of the considered algorithms is uniformly better than the other but for correlated dictionaries, guaranteed exact recovery may be obtained after fewer iterations for OLS than for OMP.