Seismic fault detection with iterative deep learning
Ruoshui Zhou, Yufei Cai, Fucai Yu, Guangmin Hu · 2019
Due to the strong nonlinear fitting ability and feature extraction ability of deep learning, a large number of researches choose deep neural networks for seismic fault detection and have got good results. However, the fault detection results lack the constraint of continuity. Inspired by the process of human learning knowledge: When learning new knowledge, humans often use the experience accumulated in the past to deepen the understanding of knowledge, but the experience may contain error information, therefore we need to constantly revise our understanding with examples until we master knowledge. We proposes iterative deep learning: in order to improve continuity, we use automatic image processing with geological expert experience to correct the fault detection results of deep learning. The complexity of the actual geological situation makes some differences between the image processing results and the actual situation. Therefore, we iteratively corrected the model through seismic data examples and image processing results. The experimental results demonstrate the effectiveness of our method. Presentation Date: Tuesday, September 17, 2019 Session Start Time: 1:50 PM Presentation Time: 3:55 PM Location: Poster Station 2 Presentation Type: Poster