Micro-Doppler (m-D) is a useful feature of radar target, as an important characteristic for radar target detection, recognition, and classification. However, since the form of m-D signals is always multi-component, the signals usually need to be separated before feature extraction. A method for separation the m-D signals is proposed based on direction pattern. In this paper, we first obtain the Time Frequency Distribution (TFD) of the target echo by Short-Time Fourier Transformer (STFT). Then the local maxima algorithm is applied for the instantaneous frequency (IF) track coherence. Finally, the direction patter algorithm is used to separate the m-D signals. Simulation results indicate that the proposed method is effective.