基于多尺度EMD的激光雷达信号分段去噪

Lidar signal subsection denoising based on multi-dimension EMD

介绍了一种新的信号去噪方法。针对激光雷达信号和 EMD 信号分解的特点,结合统计检测理论,利用中心极限定理和 Grubbs 判据对信号强度起伏的随机性进行检测。检测出的信号强突变部分视为信号的发展趋势保留不参与 EMD 分解,其余部分则使用 EMD 方法分解。对分段处理的信号用半软阈值限幅的办法重构;重构信号可能出现的 pseudo-Gibbs 现象,采用平移不变量去噪原理处理。实测信号的处理结果表明了这种去噪方法的有效性,即能取得较好的去噪效果,又能较好的保留信号的突变信息和强突变信息.

A new signal denoising method is introduced.According to the characteristics of lidar signal and EMD signal decomposition,with statistical check theory,Central limit theorem and Grubbs rule,the randomness of intensity fluctuations of lidar signal is checked.The part of high abrupt change in signal could be regarded as trend of development of signal,which is not decomposed by EMD.The remainder signal is decomposed.The part decomposition signal is reconstructed by Semi-Soft threshold limiting method.Pseudo-Gibbs phenomena may be appeared in the reconstructed signal;it could be eliminated by the application of Translation-linvariant denoising theory.The experiment shows the effectivity of this method,and it can remove noise from signal effectively and keep abrupt change and high abrupt change well in signal.