A multiple choice greedy algorithm for resolution improvement of spectral lines identification

The paper analyses the possibility to improve the resolution of spectral lines identification based on zero-padding Discrete Fourier Transform (DFT), by using a sparse vector estimation technique. A variant of the Orthogonal Matching Pursuit (OMP) algorithm called `Multiple Choice OMP (MCOMP)´ is proposed. This algorithm considers the difficulty arisen by the errors generated by the maximum coherence criterion in atoms selection, because of the high degree of dictionary coherence. It can be applied for a limited number of frequencies, with total guarantees or probabilistic guarantees.