Abstract: Becauce of interruption contained in a Lidar echo signal, mode mixing is often generated when using EMD (Empirical Mode Decomposition) to denoise such a signal. It lead to that it can't remove the noise from useful signal easily, and make the denoising effect so worse. In order to solve this problem, a combinational algorithm was presented which combine the morphological filtering and EMD together. Firstly, an adaptive multi-scale morphological filter was used to dispose the signal and remove the interruption, as a preliminary treatment, then used EMD for denoising. At last, a simulated signal denoising experiment and a real Lidar echo signal denoising experiment were done, the results showed that SNR increased by 8.89 dB and RMSE reduced by 0.051 4 compared with using EMD to denoise directly in the former experiment, the mean-SNR after 6 km increased by 3.356 4 dB in the later. This combinational algorithm can restrain mode mixing effectively, and has a better denoising effect and application prospects.