<![CDATA[Accurate segmentation of salt bodies in seismic images is a critical task in subsurface exploration, as salt structures often act as traps for hydrocarbons. Traditional manual and rule-based methods are time-consuming and prone to inaccuracies due to the complex morphology and low contrast of salt boundaries. In this study, we propose a robust multi-scale deep neural network framework designed to enhance salt body segmentation in seismic datasets. The framework leverages a multi-scale encoder-decoder architecture integrated with Atrous Spatial Pyramid Pooling (ASPP) and attention mechanisms to effectively capture both global context and fine-grained structural details. Evaluated on the publicly available TGS Salt Identification Challenge dataset, the proposed model outperforms several state-of-the-art baselines in terms of Intersection over Union (IoU), Dice coefficient, and overall segmentation accuracy. The results demonstrate the frameworkâs effectiveness in accurately delineating salt regions, even in the presence of noisy or ambiguous seismic data, offering a reliable tool for aiding geophysical interpretation and exploration.]]>
Copyrights © 2025
