Abstract:

In order to improve the signal to noise ratio and fidelity of seismic data by the compressed sensing reconstruction method, an algorithm of compressed sensing reconstruction of seismic data based on curvelet transform is proposed. A seismic data reconstruction model is built, the energy and entropy distribution characteristics of multi-scales seismic data based on sparse representation of curvelet are analyzed, the computational complexity of the random observation is reduced with the block compressed sensing technology, with the change of information entropy, an adapt bivariate shrinkage threshold iterative reconstruction method is designed based on the correlation between the high frequency region of multi-scales in curvelet domain. Experimental results show that the proposed algorithm gains above 1. 5 dB, and has better ability to maintain the detail information than the other algorithms mentioned, under the same sampling rate.

Key words: curvelet transform, compressed sensing, seismic data reconstruction, sparse representation, adaptive iterative threshold shrinkage