基于经验模态分解的曲波阈值去噪方法

针对曲波阈值去噪方法阈值选取单一造成的有效信号损失或随机噪声压制不完全的问题,笔者提出了一种基于经验模态分解的曲波阈值去噪方法。该方法首先对带噪信号进行经验模态分解得到一系列固有模态函数,根据每个固有模态函数所含噪声强弱的不同,选取不同的阈值分别对分解得到的含噪固有模态函数进行曲波阈值降噪处理;最后将去噪后的固有模态函数与不含噪声的固有模态函数进行信号重构得到最终压噪的结果。由于引入经验模态分解,对分解得到的不同含噪程度的固有模态函数,选取不同的阈值进行处理,这样能够有效减小直接曲波阈值方法阈值选取单一产生的问题。模型和实际数据试算表明,该方法在提高数据信噪比的同时,能够有效地保留有效信号,是一种相对保幅的去噪方法。

A Method of Curvelet Threshold Denoising Based on Empirical Mode Decomposition

In order to avoid the loss of effective signal or incomplete suppression of random noise because of single threshold selection in the curvelet threshold denoising method,a new curvelet threshold denoising method based on empirical mode decomposition(EMD) was proposed.Firstly,noise signal was decomposed into a series of intrinsic mode functions(IMF) by EMD.Then according to the distributing level of noise in each IMF,different thresholds were chosen to process the IMFs with noise.Finally,denoised signal was obtained by reconstructing the denoised IMFs and the IMFs without noise.Thanks to the introduction of EMD,different threshould can be chosen to apply to IMFs with various degree noise.In this way,the method can overcome the shortcomings of single threshold selection of Curvelet threshold denoising method.Applying this method to synthetical and real field data sets indicate that it can improve the signal-to-noise ratio,meanwhile,also greatly maintain the effective signal.The method is proved to be an effective and preserved-amplitude denoising way.