Abstract: In the process of transmitting to receiving a full-waveform LiDAR signal, noise is generated due to the influence of the propagation medium, the scanning distance, the nature of the object and other factors.Aiming at the problem,this paper has put forward a modified denoising method based on empirical mode decomposition (EMD), permutation entropy (PE) and wavelet threshold method (WTM). Firstly, the initial echo signal is decomposed into several intrinsic mode functions (IMF) and then the PE value corresponding to each IMF is computed. Secondly, the PE value is utilized to form the wavelet threshold value, and new threshold function is constructed. Finally, after being denoised by corresponding WTM, all the IMFs are added and the denoised signal is obtained. Features of EMD, PE and WTM are took advantage of, which enables the new method to deal with noised echo signal adaptively. By making data simulation and measured data experiment, the new method is compared with other methods. Normalized indexes and comprehensive index based on signal-noise ratio, waveform relevance, root mean squared error and smoothness indicate that, the new method increase the normalized signal-noise ratio index by 10% to 20%, while the other indexes are advanced by 5% to 40%. It can be concluded that the new method put forward in this article has better effect and is able to deal with echo signal with different levels of noise during use of full-waveform LiDAR.

Key words: full-waveform LiDAR, signal denoising, empirical mode decomposition, permutation entropy, wavelet threshold method