摘要
由于受经济成本、地质条件等因素限制,地震采集数据一般为欠采样数据且含有噪声,将对数据处理和地质解释产生严重影响。为此,基于数据驱动紧框架(data-driventightframe,DDTF)理论,研究了三维地震数据去噪和重建问题。DDTF理论限定学习字典为一组平移不变的冗余小波紧框架,通过进一步控制字典的自由度,使DDTF算法拥有良好的鲁棒性,并且利用小波紧框架完美的重构特性,可更好地保留数据的精细特征。仿真实验和实际数据应用结果表明:DDTF算法对结构简单的三维合成地震数据及结构复杂的实际三维地震数据都具有良好的去噪和重建效果,但计算效率较低,还需进一步改进;曲波变换对实际数据的去噪和重建效果较差;块匹配四维协同滤波的去噪和重建结果过于光滑,会丢失一些结构特征。
Constrained by economic and geological conditions and other factors,generally seismic data are undersampled and noises are serious,which severely impact data processing and interpretation.We studied how to denoise and reconstruct 3D seismic data on a data-driven tight frame(DDTF).The DDTF theory defines the learning dictionary to be a translation-invariant redundant wavelet tight frame.By further controlling the degree of freedom of the dictionary,it makes the DDTF algorithm have good robustness,and the fine features of seismic data can be better preserved by making use of perfect reconstruction on the wavelet tight frame.Model and real data have proved that the DDTF algorithm works well in denoising and reconstructing 3D synthetic seismic data with simple structures and real 3D seismic data with complex structures,but the computational efficiency is low and needs to be further improved.In addition,curvelet transform is less effective for denoising and reconstructing real data;the denoised and reconstructed results from block matched four-dimensional cooperative filtering are too smooth to preserve some structural features.
作者
陈杰
牛聪
李勇
黄饶
陈力鑫
马泽川
CHEN Jie;NIU Cong;LI Yong;HUANG Rao;CHEN Lixin;MA Zechuan(School of Geophysics,Chengdu University of Technology,Chengdu,Sichuan 610059,China; CNOOC Research Institute Co.Ltd.,Beijing 100027,China;State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation(Chengdu University of Technology),Chengdu,Sichuan 610059,China)
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2020年第4期725-732,699,共9页
Oil Geophysical Prospecting
基金
国家科技重大专项“大型油气田及煤层气开发重大专项——时频聚集流体识别方法研究”(2016ZX05026001-004)
四川省重点研发计划项目“基于人工智能去噪的页岩气储层微裂缝识别研究”(2020YFG0157)联合资助
关键词
三维地震数据
去噪
重建
数据驱动紧框架
3Dseismic data
denoising
reconstruction
data-driven tight frame
