摘要
随着油气勘探的发展,采集的数据规模与复杂度越来越大,对这些数据进行重建的精度与效率影响到后续地震资料的处理效果。常用于地震数据重建的压缩感知理论与重建算法各有精度与效率的优势,因此对于大规模、复杂地震数据,综合考虑重建精度与计算时间,提出了一种基于压缩感知理论和L1范数谱投影梯度算法(SPGL1)的地震数据重建方法。首先根据地震数据的缺失情况选择采样矩阵,然后在contourlet域中采用L1范数谱投影梯度算法重建缺失的稀疏系数,最后进行contourlet反变换实现地震数据的重建。合成地震数据实验结果表明,基于压缩感知和L1范数谱投影梯度算法重建的地震数据精度较好,计算效率高。通过实际地震资料处理,对比了相同稀疏变换基情况下常用的贪婪算法中的正交匹配追踪(OMP)、梯度投影稀疏重建算法(GPSR)及L1范数谱投影梯度算法(SPGL1)的应用效果,发现基于压缩感知的L1范数谱投影梯度算法鲁棒性较好,受噪声影响小,重建精度高,并且兼顾了计算效率的需求。
With the development of oil and gas exploration,the scale and complexity of collected data are increasing. The reconstruction of seismic missing data is essential for subsequent data processing.Reconstruction algorithms based on compressive sens ing are accurate and efficient. Here,we proposed a seismic data reconstruction method, based on a spectral projection gradient L1 algorithm (SPGL1) and on compressive sensing, which can be applied to large-scale and complex seismic data. First, a sampling matrix was selected according to the missing data.Then, the missing sparse coefficients were reconstructed using the SPGL1 in the contourlet domain.Finally,he contourlet inverse transform was used to reconstruct the seismic data.Tests on synthetic and field data demonstrated the superiority of the proposed method over traditional methods: it provided higher accuracy and efficiency. Based on the contourlet transform,we could conclude that the SPGL1 is more robust than OMP and the gradient projection algo rithm GPSR in the processing of noisy data.
作者
兰天维
韩立国
张良
LAN Tianwei;HAN Liguo;ZHANG Liang(College of Geo-exploratiom Science and Technology,Jilin University Changchun 130026,China)
出处
《石油物探》
EI
CSCD
北大核心
2019年第2期219-228,244,共11页
Geophysical Prospecting For Petroleum
基金
国家重点研发计划课题"天然气水合物高精度三维地震数据处理和成像技术研究"(2017YFC0307405)资助~~
关键词
压缩感知
测量矩阵
CONTOURLET变换
地震数据重建
贪婪算法
绕射波
SPGL1
compressive sensing
measurement matrix
contourlet transform
reconstruction of seismic data
greedy algorithm
dif fraction waves
spectral projection gradient L1 algorithm(SPGL1)
