In order to rapidly achieve accurate direction‐of‐arrival (DOA) estimation of multiple targets using a single measurement vector (i.e., array signal vector), we propose a new efficient sparse representation algorithm based on the orthogonal matching pursuit (OMP) using a step‐by‐step search approach. Regardless of the high mutual coherence of redundant dictionary (i.e., array manifold matrix), the optimal sparse solutions for DOAs of multiple targets can be yielded with a relatively low computational cost, as compared to the iterative local searching OMP method that is up‐to‐date. Several simulation and experimental results demonstrated that the proposed scheme is much more effective than the existing subspace‐based algorithms in a noisy environment and other sparse representation algorithms, even in actual measurement cases.