Curvelet threshold denoising joint with empirical mode decomposition

Abstract

The separation of signal and noise is an important issue in seismic data processing. By noise we refer to the incoherent noise that exists in the data. In this paper we use empirical mode decomposition (EMD) and recently introduced curvelet transform for suppression of random noise. With the aid of EMD, we firstly decompose the data with noise into a series of intrinsic mode function (IMF) profiles, then transform the IMF profiles into curvelet domain respectively and choose different thresholds to process them in view of the difference of noise in each IMF profiles. In this way, we can protect the effective signal as much as possible while suppressing incoherent noise.