摘要
作为一种非自适应的多尺度、多方向性几何分析方法,曲波变换能够近乎最优地表示含奇异点的高维曲线,地震数据在曲波域有更好的稀疏表达。对于多炮地震数据,三维曲波变换能够成功地实现信号分离,达到去除随机噪声的目的。阐述了三维曲波变换的基本原理;将三维曲波变换与阈值迭代法结合起来,对不同信噪比的模拟数据和实际地震资料进行去噪处理,并与传统的中值滤波法、F-X反褶积法及二维曲波阈值迭代法处理结果进行量化对比。结果表明,基于三维曲波变换的地震数据去噪方法不仅去除噪声能力更强,而且能够保护有用信息,是一种有效的多炮地震数据去噪方法。
As a non adaptive,multi-scale and mult-directionality geometrical analysis method,Curvelet transform can almost optimally describe high dimensional curve with singularity,and seismic data can be better sparsely expressed in curvelet domain.For multi-dimensional seismic data,by 3D Curvelet transform we can successfully achieve signal separation and eliminating random noise.In this paper,the basic theory of 3D Curvelet transform is elaborated,and the denoising method based on 3D Curvelet transform combined with the threshold iterative algorithm is proposed.The 3D Curvelet thresholding method is applied to denoise processing on synthetic data and practical seismic data with different SNR,and the processing results are quantitatively compared with the ones from traditional denoising methods,such as median filter,F-X deconvolution and 2D Curvelet thresholding.Comparative analysis shows that the 3D Curvelet thresholding method can remove the random noise effectively and be able to protect useful information.It＇s an effective denoising method for multi shots seismic data.
出处
《石油物探》
EI
CSCD
北大核心
2014年第3期313-323,共11页
Geophysical Prospecting For Petroleum
基金
国家科技重大专项(2011ZX05023-005-008)资助
关键词
三维曲波变换
随机噪声压制
阈值迭代法
L1范数
高信噪比
3D Curvelet transform
random noise eliminating
threshold iterative algorithm
L1 norm
high SNR
