A modified orthogonal matching algorithm using correlation coefficient for compressed sensing

This paper presents a modified orthogonal matching pursuit (OMP) algorithm for compressed sensing (CS). Compared with the standard OMP algorithm, the most innovation of this algorithm is its improvement on the reconstruction probability of the sparse signal, and the basic idea is that the support set is estimated using correlation coefficient because the correlation coefficient can be viewed as a normalized matching criterion. In the standard OMP algorithm, the inner product is used for estimating the support set of the sparse signal, which may generate the wrong coordinates because the inner product couldn't guarantee the expected column of sensing matrix matches the measurement vector very best. However, the proposed algorithm is able to demonstrate a better performance on estimating the support set to some extent. From the simulation results, the proposed algorithm outperforms the standard OMP algorithm.