Seismic Scattering Wave Field Imaging Method Based on Convolution Neural Network and Equivalent Training Model

The inversion of scattered waves in seismic exploration is a hot and difficult problem related fields. We propose hypothesis: the local wave field near each point on profile has relationship with minimum distance between that scatterer, this can be recognized by convolutional neural network (CNN). Based this, designed CNN steadily classify identify wavefield point, realize imaging scatterer. new method based equivalent training model. transforms optimal scatterer into design Through model constructed parameterized method, two different layered random cave media models are well realized, Marmousi2 synthesis data also realized robustly. use Bayesian discriminant improves resolution results helps interpreters quickly accurately locate scatterers.