Scholarlink

<p>Seismic image segmentation is important in several geological feature identifications to accurately delineate structures and regions of interest. This paper presents a comprehensive study of the application of the U-Net architecture for seismic image segmentation of diverse chaotic seismic reflections, focusing on its effectiveness in interpreting and analyzing the data. We provide a detailed explanation of our proposed U-Net architecture, the implementation specifications, and demonstrate its performance on a dataset of seismic images. The training data set has been generated and segmented manually from the F3 seismic data set from the North Sea of the Netherlands. Accuracy and Dice coefficient are used as evaluation metrics, and U-Net architectures with various hyperparameters are compared. Performance measures show an average accuracy of 95.6% and Dice coefficient of 0.85 for the predicted complex and chaotic segmented region in seismic images.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Application of U-Net architecture for segmentation of complex chaotic reflections from seismic images

  • 期刊论文

  • 发表日期：

    2026年5月13日

查看全文
  • Avik Roy,
  • Anuja Arora

摘要

Seismic image segmentation is important in several geological feature identifications to accurately delineate structures and regions of interest. This paper presents a comprehensive study of the application of the U-Net architecture for seismic image segmentation of diverse chaotic seismic reflections, focusing on its effectiveness in interpreting and analyzing the data. We provide a detailed explanation of our proposed U-Net architecture, the implementation specifications, and demonstrate its performance on a dataset of seismic images. The training data set has been generated and segmented manually from the F3 seismic data set from the North Sea of the Netherlands. Accuracy and Dice coefficient are used as evaluation metrics, and U-Net architectures with various hyperparameters are compared. Performance measures show an average accuracy of 95.6% and Dice coefficient of 0.85 for the predicted complex and chaotic segmented region in seismic images.