A segmentation singular value decomposition (SVD)-lifting wavelet transform (LWT) denoising algorithm based on ensemble empirical mode decomposition (EEMD) was proposed to better suppress noise in an atmospheric lidar return signal. The EEMD method is used to distinguish inherent modal functions (IMFs) of the noise and signal, and remove the IMF with noise as its main component. Moreover, the SVD-LWT method is adopted to remove the noise in the IMF component containing the signal and thus finely extract the signal. The simulated Bumps signal with different sequences of Gaussian white noise was denoised, and the denoising effect of the EEMD-SVD-LWT algorithm was compared with the effects of the wavelet soft threshold, EEMD (correlation coefficient), and EEMD (difference value) methods. Simulation shows that the denoising effect of the EEMD-SVD-LWT algorithm was best. The EEMD-SVD-LWT algorithm was also used to denoise practical lidar signals and was better than that achieved with the other methods.

提出了一种基于整体经验模态分解（EEMD）的分段奇异值分解-提升小波变换（LWT）降噪算法，以更好地抑制大气激光雷达返回信号中的噪声。EEMD方法用于区分噪声和信号的固有模态函数（IMF），并删除以噪声为主要成分的IMF。此外，采用SVD-LWT方法去除包含信号的IMF分量中的噪声，从而精细地提取信号。对具有不同高斯白噪声序列的模拟Bumps信号进行去噪，并将EEMD-SVD-LWT算法的去噪效果与小波软阈值，EEMD（相关系数）和EEMD（差分值）方法的效果进行比较。仿真表明，EEMD-SVD-LWT算法的去噪效果最好。EEMD-SVD-LWT算法也被用于对实际的激光雷达信号进行降噪，并且优于其他方法。