摘要
经验模态分解(Empirical Mode Decomposition,EMD)是一种信号的时频分析方法,该方法在不需要先验知识的条件下,可以将非平稳、非线性信号,依据信号的特征,自适应的分解为多个本征模态函数(Intrinsic Mode Function,IMF)之和,得到高的频率分辨率。然而,一般的去噪方法是将所选择的高频IMF部分取不同的阈值进行滤波或者是直接置为零重构后实现信号的去噪,很显然这会造成高频部分有用信号的损失。1D全变分(Total Variation,TV)是一种有效的信号去噪方法,能够非常好的保护信号边缘信息,但有时也会把噪声当作边缘信息,出现虚假边缘现象。因此,基于EMD和1D-TV的优点提出了一种新的去噪方法,根据对实际金属矿床地震信号处理的结果表明,该算法能有效的消除地震信号中的噪声,并能有效保护地震信号边缘构造信息。
Empirical Mode Decomposition（EMD） is a kind of time-frequency analysis method of signal. It does not require a priori knowledge,and can self-adaptively decompose non-stationary and nonlinear signals into multi-scale Intrinsic Mode Functions（IFM） in terms of the nature of the signals,which leads to the high frequency resolution. However,common de-noising method filters the high frequency part chose from IMF from different threshold or realizes signal de-noising by setting it to zero directly and refactoring it,which will lead the obvious loss of useful signal of high frequency part. 1D Total Variation（TV）,is a effective signal de-noising method,can preserve edges data better,but sometimes it may take the noise as edge information,so as to appear the false edge. Here,We construct a new noise attenuation algorithm by combining EMD with 1D total variation de-noising method,the process result of the metal deposit seismic signal shows,the algorithm can effectively separate seismic signals form random noise and keep the edge structure information of the seismic data well.
出处
《四川理工学院学报（自然科学版）》
CAS
2014年第3期29-33,共5页
Journal of Sichuan University of Science & Engineering(Natural Science Edition)
基金
国家863计划项目(2008AA121103)
中国地质调查局项目(1212011120226)
