Batch algorithms of matching pursuit and orthogonal matching pursuit with applications to compressed sensing

Batch algorithms of matching pursuit (MP) and orthogonal matching pursuit (OMP) are proposed in this paper. In both algorithms, the original iteration procedures are modified in the following way. Instead of finding a single best-matched atom in each iteration, we find a number of best-matched atoms to speed up the convergence, - a batch version. Then optimized coefficients are computed based on these atoms. Numerical simulations in the application to compressed sensing show that the proposed algorithms are much faster than the original ones, while similar reconstruction precision is obtained.