In the past few years, deep learning methods have been used in many industries and achieved great success. In this work, we try to enhance seismic data’s resolution by utilizing the deep learning which can learn features in different level and merge them to recover the missing resolution. Firstly, we introduce the generative adversarial network to seismic data processing and use it to enhance the resolution of seismic data. For a 3D field seismic dataset, we use one typical method, the adaptive bandwidth extension in the continuous wavelet transform domain, to enhance the resolution of the dataset. We choose 64 percent of the dataset as our network’s training dataset, while the corresponding 64 percent of the resolution enhancing result are adopted as the label dataset. After well training, we send the rest 36 percent of field seismic dataset into our network and compare the resolution enhancement results generated by our network with the result generated by the conventional method. The comparisons show that our network not only can obtain a comparable result with conventional method but also recover more tiny reflection than the conventional method.
Presentation Date: Tuesday, September 17, 2019
Session Start Time: 9:20 AM
Presentation Time: 10:10 AM
Location: Poster Station 2
Presentation Type: Poster
