Seismic data reconstruction based on K-SVD dictionary learning under compressive sensing framework
Zhou Yatong1,2, Wang Lili1, Pu Qingshan3
1. School of Information Engineering, Hebei University of Technology, Tianjin 300401, China;
2. School of Mathematical Science, Peking University, Beijing 100871, China;
3. BGP International, BGP Inc., CNPC, Zhuozhou, Hebei 072751, China
Abstract：A novel seismic data reconstruction method combined with compressive sensing and K-SVD dictionary learning is presented in this paper to deal with incomplete or irregular seismic data due to acquisition cost or environmental factors. First we obtain an over complete dictionary by training K-SVD dictionary learning on a large number of seismic data samples, then we introduce the sampling matrix of irregular seismic data as a measurement matrix. After that, the seismic data is reconstructed by the regularized orthogonal matching pursuit algorithm. Unlike conventional reconstruction algorithms based on curvelet transform or Fourier transform using single orthogonal basis, the over complete dictionary introduced in this paper extracts adaptively feature through the training samples, and can adaptively select transform basis according to the characteristics of processed data itself. The over complete dictionary provides a flexible way to seismic data sparse extension, and could lead to a better reconstruction result. Experiments on the synthetic seismic data and real marine seismic data have verified the feasibility and efficiency of the method.
周亚同, 王丽莉, 蒲青山. 压缩感知框架下基于K-奇异值分解字典学习的地震数据重建[J]. 石油地球物理勘探, 2014, 49(4): 652-660.
Zhou Yatong, Wang Lili, Pu Qingshan. Seismic data reconstruction based on K-SVD dictionary learning under compressive sensing framework. OGP, 2014, 49(4): 652-660.
