Elena Klochikhina | Deep Learning for Migration Artefacts Attenuation
10 December | Session: ML in Seismic Processing 1 | 14:00 - 15:00 (CET), 2nd speaker. Deep Learning for Migration Artefacts Attenuation
Authors: Elena Klochikhina, Sergey Frolov, Nizar Chemingui (PGS). Click the image below for the abstract.
Elena Klochikhina presents a machine learning-based approach for attenuating migration artifacts from seismic sections. A U-net-like convolutional neural network was trained on synthetic data to remove uncancelled migration isochrones. A representative dataset was collected for the model training. Application of de-noising with the neural network on two different field data examples shows the capability of the method to preserve geological structures in the images while attenuating the noise.