Title

Random noise attenuation based on compressive sensing and TV rule

单位

(1.中国石油大学(北京)油气资源与探测国家重点实验室,北京昌平102249;2.中国石油大学CNPC物探重点实验室,北京昌平102249)

Organization

(1.State Key Laboratory of Petroleum Resources and Prospecting,China University of Petroleum,Beijing 102249,China; 2.CNPC Key Lab of Geophysical Prospecting,China University of Petroleum,Beijing 102249,China)

摘要

考虑到传统曲波变换阈值法难以获得理想的去噪效果,提出一种基于压缩感知和全变差(Total Variation,TV)准则约束的曲波变换地震资料去噪方法。该方法将传统的去噪问题转化成TV准则约束下的最优化问题,通过求取最优解重构原始信号,从而实现对地震资料的去噪处理。理论模型数据测试结果表明,该方法在地震资料去噪过程中可以有效压制同相轴边缘附近产生的不光滑畸变,相比传统的曲波变换阈值法能够获得更好的去噪效果;与F-X反褶积相比,该方法不仅能够有效压制地震资料中的随机噪声,提高地震资料的信噪比,而且能够较好地保护有效信号。某工区实际地震资料去噪结果进一步验证了上述方法的有效性。

Abstract

Owing to the undesired noise attenuation effect of conventional threshold method based on Curvelet transform,a seismic data denoising approach based on compressive sensing and Total Variation (TV ) rule is proposed in this paper.It changes noise attenuation into the optimization problem under the constraint of TV rule,the original signal is reconstructed by optimum solution and thereby the denosing results of seismic data are obtained.Theoretical numerical model test shows the proposed method can not only effectively suppress the distortion near the edges of seismic events but also achieve better results compared with conventional Curvelet transform threshold method.Moreover,compared with F-X deconvolution,this approach can attenuate the random noise effectively and preserve effective signal with an improvement of signal to noise ratio in seismic data.Finally,the denoising results of actual seismic data prove the effectiveness of the proposed method further.

关键词：

压缩感知; TV准则; 曲波变换; 随机噪声; 噪声衰减;

Keywords:

compressive sensing; TV norm; Curvelet transform; random noise; noise attenuation;

基金项目

国家科技重大专项(2011ZX05024-001-01)和国家自然科学基金项目(41140033)共同资助。

DOI

10.3969/j.issn.1000-1441.2015.02.009

基于压缩感知和TV准则约束的地震资料去噪