Adaptive noise attenuation of seismic images based on singular value decomposition and texture direction detection

摘要

Singular value decomposition (SVD) is an efficient tool for the separation of signal and noise subspaces. When it is used to process seismic images, SVD can enhance the signal-to-noise ratio (SNR) of horizontal events effectively. In this paper, an adaptive SVD filter is proposed to enhance the non-horizontal events by detection of seismic image texture direction and then horizontal alignment of the estimated dip through data rotation. The features derived from the co-occurrence matrix are used to estimate the texture direction. The SVD filter parameter is adapted according to the ratio of the stacking energy along the detected direction and the energy of the image. Coherent noise events are recognized by their directions, which are different from the directions of signal events in general, and are first attenuated by high-rank approximation. Then, the signal events are enhanced by low-rank approximation.