Research On Seismic Signal Processing Based On Sparse Representaion And Compressive Sensing And Its Application

With the more complicated application environment of seismic exploration, the seismic data were always contaminated by noise and other sort of stuffs, so that the quality of seismic data could not be used to predict and interpret. How to recover all these contamination is a new challenge to the seismic data process flow. Fortunately, the seismic data are sparse in frequency domain. This feature with the newly mentioned compressive sensing theory could be used to help us to raise the quality of seismic data efficiency.With the deep learning and understanding about sparse representation and general compressive sensing framework, This thesis discussed about the seismic data denoise method, recovery method and the implement of super resolution of seismic data who are both realized by sparse feature and compressive sensing method. The work this thesis covered are listed below:This thesis proposed a noise damping algorithm combined with shearlet sparse representation and anisotropy diffusion procedure. It could damp the noise and enhance the seismic signal simultaneously, in order to make the events in the seismic data obvious. And we did some simulation and test to prove the efficiency and performance.Beyond that, we used series promote sparse algorithms to achieve seismic recovery.Based on the analysis of the model of seismic hiatus, we tested and analyzed the advantages and shorts of these algorithms, and proposed the different field of application of these algorithms in seismic recovery problem.At last, this thesis proposed a strategy of achieving the super resolution of seismic data. We combined the shearlet sparse representation, noise damping, and seismic recovery technique to realize a two-step method of super resolution which is to raise the time and space resolution respectively. And we have proved and tested this strategy with both theoretic and real seismic data.