next up previous [pdf]

Next: Acknowledgments Up: Chen & Fomel: EMD-seislet Previous: Application of noise attenuation


We have proposed a novel approach for sparsifing seismic data: EMD-seislet transform. The approach relies on the ability of empirical mode decomposition (EMD) in the $ f-x$ domain to provide smoothly non-stationary data, which we use in the following 1D non-stationary seislet transform. When applied to seismic data with multiple conflicting slopes, EMD-seislet is remarkably sparse. A real data example shows an excellent performance of the EMD-seislet transform in attenuating random noise. However, the large computational cost required by the EMD algorithm requires a careful design of the parallel computing framework in the future research.