Abstract: Aiming at the problem that the traditional direction of arrival（DOA） algorithm had low estimation accuracy under the conditions of small snapshot number, low signal-to-noise ratio（SNR） and coherence signals, a DOA estimation based on detour matching pursuit algorithm was proposed. The space was divided into equal angles, and the super-complete redundant dictionary was built. The sparse signals were reconstructed by the detouring matching pursuit algorithm, and the non-zero elements in the reconstructed signals were converted into the corresponding direction angles according to its position numbers, that was the correct DOA estimation. The simulation results show that the DOA estimation algorithm based on detouring matching pursuit algorithm can still get effective estimation value and has good robustness under the conditions of single snapshot and strong noise.