Random noise attenuation by 3D Multi-directional vector median filter

Abstract

Noise attenuation is an important data processing step in exploration geophysics, which is used to extract useful information or improve the signal to noise ratio (SNR) from the raw measurements. The vector median filter (VMF), as an extension of the scalar median filter (SMF), can efficiently suppress random noise and preserve discontinuities of the spreading structures within the data, but it faces questions of how to define a minimum distance and what appropriate additional constraints should be introduced for a reasonable unique median-output vector. Additionally, to avoid uncertainty in data processing using the 2D-VMF method and to more reasonably describe the 3D seismic cube, this study introduces a 3D multi-directional VMF. The new 3D-MDVMF method primarily consists of two steps: 1) detecting the main local dip using some trial angles with a similarity judgement and 2) applying the filter with the constraints for noise suppression and structure enhancement. Synthetic simulated data and marine data are tested. The results show that the proposed method can enhance local dip detection, separate the random noise, and well preserve discontinuities of the structures. This method might be significant for general noise attenuation for data containing spreading structures. KeywordsNoise attenuationMedian filterSeismic dataMulti-directional operator