Seismic Image Completion Method Based on Information Entropy Norm

Abstract

The invention discloses a seismic image completion method based on the information entropy norm, which is applied in the field of geophysical seismic data processing, and the low-rank matrix recovered by the method based on matrix rank reduction is not For the real low-rank problem, the present invention replaces the nuclear norm used by the original SVD with an objective function based on the information entropy norm, and then regularizes the rank minimization term of the objective function with the information entropy norm, and finally iterates the ADMM estimation The method is used to approximate the solution to obtain a low-rank matrix, so as to obtain a robust restoration of missing images.

Description

technical field [0001] The invention belongs to the field of geophysical seismic data processing, and in particular relates to a seismic data complementing technology. Background technique [0002] In the actual geophysical seismic exploration and acquisition, there are two main reasons for the lack of seismic data: (1) The surface environment of the work area (such as rivers, ponds, lakes, etc.), surface obstacles (roads, houses, bridges, etc.) ) and other influences, the layout of the geophone is often irregular; (2) due to the poor coupling effect between the geophone and the ground surface, environmental interference or the instrument itself will produce waste tracks and waste guns. With the development of seismic exploration technology, the volume of collected data is increasing. On the one hand, we hope that densely sampled data can obtain more accurate underground structures. On the other hand, considering the cost of exploration, we need to sparsely sample in the spa...

AI Technical Summary

Problems solved by technology

[0004] The core of the method based on matrix rank reduction is to use the L1 norm of the singular value vector of the matrix to describe the low rank of seismic data, but the L1 norm description has the problem of insufficient sparse description: for example, for a=[4,4,4 ,0,0] and b=[10,1,1,0,0] two singular value vectors, according to the definition of the L1 norm, it is impossible to distinguish accurately, in fact, the sparsity of the singular value b vector is much higher Due to the singular value a vector, the limitation of the L1 norm leads to the low-rank matrix recovered by the method based on matrix reduction rank is not really low rank

Images