基于空间自适应高阶全变分的地震数据去噪模型  

Seismic data denoising model based on spatially adaptive high-order total variation

作　　者：李江萍 王德华[1] 乔增强 张丽丽[2] 刘乃豪 LI Jiangping;WANG Dehua;QIAO Zengqiang;ZHANG Lili;LIU Naihao(School of Sciences,Xi’an Technological University,Xi’an 710021,China;School of Freshmen,Xi’an Technological University,Xi’an 710021,China;School of Information and Communication Engineering,Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]西安工业大学基础学院,西安710021 [2]西安工业大学新生院,西安710021 [3]西安交通大学信息与通信工程学院,西安710049

出　　处：《非常规油气》2025年第1期1-8,共8页Unconventional Oil & Gas

基　　金：陕西省数理基础科学研究项目“油气勘探中地震反演问题的贝叶斯深度学习方法研究”(23JSY044);教育部“春晖计划”合作科研项目“基于贝叶斯深度学习的非平稳地震记录波阻抗反演方法研究”(202200955)。

摘　　要：地震噪声压制是地震数据处理的关键环节,其结果将影响地震数据的后续处理及地质解释。针对传统全变分(TV)正则化模型容易导致阶梯效应,以及高阶全变分(HOTV)正则化模型容易丢失边缘信息的问题,提出了一种能够克服阶梯效应并保护边缘信息的地震数据去噪模型--空间自适应高阶全变差(SAHOTV)正则化模型。首先,通过差分特征值构造基于边缘检测函数的空间自适应权函数;其次,根据边缘检测函数提取的细节信息,定义空间自适应高阶全变分地震数据去噪模型;最后采用分裂Bregman迭代算法快速求解。实验结果表明:1)该方法能够提高地震数据的峰值信噪比;2)在抑制随机噪声的过程中可以显著地降低阶梯效应;3)能够较好地保留边缘及构造信息;4)该方法对有效同相轴信息损伤较小,保真度较高,去噪效果的优势明显。以上成果认识,对提高地震数据勘探的精度具有指导意义。Seismic noise suppression is a key step in seismic data processing,and its results will affect subsequent processing of seismic data and geological interpretation.This paper aims to address the issue that the traditional total variation(TV)regularization model easily causes the staircase effect,and the high-order total variation(HOTV)regularization model easily loses edge information,proposed a seismic data denoising model,spatial adaptive higher order total variation(SAHOTV),to overcome the step effect and protect the edge information.First,a spatial adaptive weight function based on the edge detection function was constructed through differential eigenvalues.Secondly,the spatial adaptive HOTV seismic data denoising model was defined based on the detailed information extracted by the edge detection function.Finally,the split Bregman iterative algorithm was used to solve the issue quickly.Experimental results showed that:1)This method can improve the peak signal-to-noise ratio of seismic data;2)It can significantly reduce the step effect in the process of suppressing random noise;3)It can better retain the edge and structural information;4)The method has less damage to the effective homodyne axis information with higher fidelity,and the denoising effect is more obvious.The above results have guiding significance for improving the accuracy of seismic data exploration.

关 键 词：地震数据去噪 随机噪声 高阶全变分 差分特征值 分裂Bregman迭代算法 

分 类 号：P631[天文地球—地质矿产勘探]