Radon域下的K-SVD字典的混采分离

Separation of blended seismic acquisition with K-SVD dictionary in Radon domain

基于地震反演来分离混采数据的方法,笔者提出用K次迭代奇异值分解(K-SVD)来更新Radon域下混叠信号中的字典原子的方法:在同步源采集和地震稀疏反演的背景下,将混合地震信号的分离视为稀疏反演问题,将共检波点域的数据表示在Radon域内,此时有效信号同相轴收敛;对数据阈值滤波后进行分块字典学习,进一步稀疏地表示地震数据;最后,固定字典,更新恢复信号和稀疏系数完成分离。模拟和实际资料处理结果表明:该方法对于混采数据的分离相对中值滤波、小波变换等更有效、分离质量明显提升,可应用于实际混叠数据中。

The authors propose an approach that can separate blended seismic acquisition data using seismic sparse inversion. Meanwhile,the proposed approach updates the dictionary atoms via K-time iterative singular value decomposition( K-SVD) in the Radon domain: the separation of blended seismic data can be taken as sparse inversion for simultaneous sources,therefore,for common receiver point data,the reflectors of the effective signal are convergent in Radon domain;after that,the filters and block dictionary learning are applied to the blended data to sparsely represent the seismic data;finally,the updated dictionary is fixed and the sparse coefficients are calculated to achieve the separation. Through the synthetic and field data experiments,it is concluded that separation method of the proposed approach is more accurate than median filter and wavelets transform,and can be applied in the separation of blended field seismic data.