Abstract

In the framework of seismic image processing, one is often interested in the restoration and enhancement of the quality of the seismic image. This process is carried out to attenuate degradation due to sensors during data acquisition or introduced by some preprocessing procedures. Random noise that can not be predicted from trace to trace can mask other seismic energy, hindering other noise attenuation processes or disguising subtle structural or stratigraphic features. Random noise attenuation by projection filtering reduces incoherent noise by eliminating data that are not linearly predictable. It can be applied on 2D or 3D seismic data, on data ordered into any domain, for example, shot gathers, cross-spreads or common offset planes either pre or poststack. Signal to noise ratio enhancement in the F-X domain has been proposed by Canales (1994) as a method for random noise attenuation. The technique is widely accepted and used in the industry. The method is very effective, easy to implement and very efficient in the computational sense.