In this paper, we tackle the problem of multiple-input multiple-output (MIMO) mmWave channel estimation in a hybrid analog digital architecture by exploiting the structure of the mmWave channel and the physical constraints of paths therein. We accomplish this by applying a novel Orthogonal Matching Pursuit (OMP) algorithm to the beamspace channel. The proposed OMP method leverages two properties of beamspace channel, which are i) relatively sparse structure, and ii) uniqueness of angle values of each path. Simulation results show that our proposed OMP method outperforms standard OMP in terms of probability of successful recovery and mean square error.