Denoising Method Based On3D Spare Transform

Seismic exploration is a geophyical exploration method aim at exploring the oil and gaseneygy,it takes advantage of the elasticity and density differences of mediumunderground,through observing and analysising the seismic wave response excited byartificial.However,due to environment conditions,consrruction,equipment and otherfactors,inevtably the seismic record acqured in the field contains different levels of noise,which seriously interfere with the image quality of the seismic record,and it is alsounfavorable to the subsequent processing and interpretation. Therefore,how to denoise betterbecomes an urgent work.With advances in computer technology recent years,the3D seismicexploration rises gradually.3D seismic exploration has attracted people due to its higheraccuracy and wider applicable conditions. At the same time, the problem of denoise for3Dseismic data has become a focus of attention.There are many kinds of noises in seismic records,denoising method also varities. Thispaper studies two denoising method for random noise suppression:One is based on thefilter,the method is take advantage of the difference between signal and noise interference inthe frequency,by the way of filtering to remove noise;the other one is besed on sparsetransform, which the method is the signal and noise projected by some sparse transformedinto a transform domain,take advantage of the difference between the two to separatedthem,to achieve the goal of noise cancellation.Firstly,this paper checked two filteralgorithms(the median filter and the F-X deconvolution filter),results showed that when theseimic record contained higher random noise,the more serious of these two methodslossed.Then this paper checked two sparse transform algorithms(curvelet transform andwavelet transform), results founded that both of the two denoising methods based on sparsetransform suppressed random noise significant more effecttly than the two filteringalgorithms.In which the method based on wavelet transfom has the disadvantange of lackingof the capacity expressing along the edges,and the method based on curvelet transform canexpress the signal almost optimally because of its multi-directional and multi-scale,it is can beseen from the comparison of the results:the SNR of seismic records increased significantlyafter denised by the curvelet tranform,and lossed little signals,which is obviouely better thanthe denoising method based on wavelet transform.Based on the excellent noise resistance of curvelet transform,we applied3D curvelettransform to the denoising problem of3D seismic data.By the calculations of theoreticalmodels and the actual data, it proved that the denoising method based on3D Curvelettransform not only suppress random noise effectively, but also can protect the signal greatly,which meets the goal of high SNR and high fidelity of seismic exploration, it is a stableeffective de-noising method for3D seismic data.Due to3D curvelet transform operated slowly,we adopted a calculation method takeplace of3D curvelet transform,which done twice2D curvelet transform along inlinedriectional and crossline driection respectively.We conducted on theoretical models and theactual data each other,then compared with3D Curvelet threshold iterative algorithm, the results proved that this alternative algorithm can suppress random noise effectively,we canuse this method as a3D Curvelet alternative when dealing with large amounts of data...