基于曲波噪声估计的三维块匹配地震资料去噪  

3D Block matching seismic data denoising based on Curvelet noise estimation

作　　者：孙成禹[1,2] 刁俊才 李文静 SUN Chengdu;DIAO Juncai;LI Wenjing(School of Geosciences,China University of Peirole-um(East China),Qingdao,Shandong 266580,China;Laboratory for Marine Mineral Resources,Qing­dao National Laboratory for Marine Science and Technology,Qingdao,Shandong 266071.China;Research Development Center»BGP Inc.,CNPC.Zhuozhou.Hebei 072751.China)

机构地区：[1]中国石油大学(华东)地球科学与技术学院,山东青岛266580 [2]青岛海洋国家实验室海洋矿产资源评价与探测技术功能实验室,山东青岛266071 [3]东方地球物理公司物探技术研究中心,河北涿州072751

出　　处：《石油地球物理勘探》2019年第6期1188-1194,I0007,共8页Oil Geophysical Prospecting

基　　金：国家自然科学基金项目“深度偏移地震数据特征剖析与深度域直接反演方法研究”(41874153);国家科技重大专项“复杂目标多尺度资料高精度处理关键技术研究”(2016ZX05006-002-003)联合资助

摘　　要：常规的三维块匹配(BM3D)算法在地震资料降噪处理中具有较好的效果,但在实际处理中因无法得到噪声先验信息,通常难以确定所需的滤波阈值等相关参数。为此,提出了一种基于曲波噪声估计的BM3D地震资料去噪方法。首先利用曲波变换估计地震资料的噪声方差,再通过改进的BM3D去噪算法自适应地选取合适的阈值参数并完成去噪处理。理论模型与实际资料的处理结果表明,所提算法与常规的BM3D去噪算法和曲波变换去噪算法相比,能在很好地去除随机噪声的同时更好地保护有效信号,且在去噪过程中对边界反射的细节信息保持较好,计算效率较高,在实际资料处理中得到良好的效果。Conventional 3D block matching(BM3D)algorithms are used for seismic data denoising.However,some parameters such as filtering threshold are difficult to be determined because of the lack of prior noise information in the practical processing.In this paper,an improved BM3D denoising method based on the Curvelet noise estimation is developed for seismic data.First the noise variance of seismic data is estimated by the Curvelet transform method.Then appropriate threshold parameters are adaptively determined.Finally the noise elimination is accurately achieved by this improved BM3D algorithm.Based on model and real data tests,the proposed algorithm can better eliminate random noise and protect signals than the conventional BM3D algorithm and Curvelet transform algorithm.Furthermore the proposed algorithm maintains most detailed information of boundary reflection and its computational efficiency is relatively high.

关 键 词：地震资料去噪 三维块匹配 噪声先验估计 曲波变换 信噪比 

分 类 号：P631[天文地球—地质矿产勘探]