We developed a supervised machine learning technique to align seismic images. Aligning seismic images is an important step in many areas of seismic processing such as time-lapse studies, tomography, and registration of P and S-wave images. This problem is especially difficult when the misalignment is large and varies rapidly and when the images are not just shifted versions of each other because they are either contaminated by noise or have different phase or frequency content. In addition, the images may be related by multidimensional vector-valued shift functions. We developed a deep learning approach for seismic image registration using convolutional neural networks (CNN). We train our CNN on images warped with known shifts and corrupted with noise, frequency, and phase perturbations. We demonstrate the promising performance of the trained CNN with synthetic data sets where the shift function is known and the images are contaminated with noise and other perturbations. The accuracy of the results obtained with our CNN is favorably compared with two other popular methods for seismic registration: windowed crosscorrelation and dynamic image warping (DIW).
Presentation Date: Tuesday, October 13, 2020
Session Start Time: 8:30 AM
Presentation Time: 11:25 AM
Location: 351D
Presentation Type: Oral