A De-noising Technology Of Seismic Signal Based On EMD And1D Total Variation

Abstract: Seismic data de-noising is a key link in seismic data processing whichmakes a direct impact on subsequent processing. Three high standards in the seismicdata processing are as follows: high signal-to-noise ratio, high-resolution andhigh-fidelity. Since the signal-to-noise ratio is the basis, to improve the signal-to-noiseratio is the primary task in the seismic data processing. With the increasinglysophisticated exploration and complex exploration conditions, in order to make moreaccurate geological interpretation, higher quality seismic data is needed, and thestudy of de-noising is more and more important.Currently, in terms of resources exploration, shallow resources has been largelydeveloped.It is necessary to seek deeper mineral resources, so people are makingefforts to seek a second prospecting space between500-2000m. Undoubtedly, it is agreat challenge for deep resource exploration techniques and methods, which alsoputs forward higher requirements for the collected field data processing. Fordecades, in order to look forward better treatment effect, people are improving theexisting treatment technologies continuously-to create a new approach, or to seekthe intrinsic link between the various methods, to get optimal processing algorithms.Firstly, the paper describes the common noise and noise characteristics of theseismic data. It gives two evaluations standards of the peak signal-to-noise ratio(PSNR) and edge retention (EPI) de-noising effect, and quantitative calculation isprovided.Empirical Mode Decomposition (EMD) is a kind of signal of time-frequencyanalysis method, it does not require a priori knowledge, can self-adaptivelydecompose non-stationary and nonlinear signals into multi-scale Intrinsic ModeFunction (IFM) in terms of the nature of the signal, that will lead to the highfrequency resolution. However, common de-noising method is will choose the highfrequency part the IMF from different threshold filtering or is set to zero directly therealization of signal de-noising. Total Variation (TV), another effective signal de-noising method, can preserve edges better, but sometimes the noise as edgeinformation, the appearance of the false edge. Here, We construct a new noiseattenuation algorithm by combining EMD with1D total variation de-noising method，which can effectively separate signals form random noise and the important detailstructures of the seismic data is kept will.At the last, a brief comparison and analysis among these de-noising methods wasmade, meanwhile some deficiencies and the ideas needed further study were given.