Scholarlink

<p>Borehole imaging is widely used in rock characteristic detection, which is of great significance for safe construction. However, the current analysis of rock strata characteristics mainly relies on geologists’ intuition and experience, which results in slow speed and low accuracy, making it difficult to meet the needs of intelligent development. A series of studies is conducted in this paper to address the problems mentioned earlier. The Deeplabv3+ network is adopted to establish a structural plane pixel recognition model, which can effectively identify structural plane pixels. Next, the thinning algorithm is adopted to extract the skeleton of the structural plane, which greatly reduces the number of structural plane pixels while maintaining the contour of the structural plane. Then, a novel method for extracting and matching fracture segments is proposed, which achieves fast and accurate identification of structural planes. Finally, the position, dip angle, and dip direction can be obtained based on the sine fitting function parameters of the structural plane. The test results show that the proposed method achieves a structural plane recognition rate of 93.7%. This method promotes the transformation of the borehole imaging system from qualitative manual analysis to intelligent quantitative analysis.</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

Research on Fast and Intelligent Recognition of Structural Planes in Borehole Images

  • 期刊论文

  • 发表日期：

    2025年9月19日

查看全文
  • Cancan Liu,
  • Xigui Zheng,
  • Haifeng Wang,
  • Niaz Muhammad Shahani,
  • Xiaowei Guo

摘要

Borehole imaging is widely used in rock characteristic detection, which is of great significance for safe construction. However, the current analysis of rock strata characteristics mainly relies on geologists’ intuition and experience, which results in slow speed and low accuracy, making it difficult to meet the needs of intelligent development. A series of studies is conducted in this paper to address the problems mentioned earlier. The Deeplabv3+ network is adopted to establish a structural plane pixel recognition model, which can effectively identify structural plane pixels. Next, the thinning algorithm is adopted to extract the skeleton of the structural plane, which greatly reduces the number of structural plane pixels while maintaining the contour of the structural plane. Then, a novel method for extracting and matching fracture segments is proposed, which achieves fast and accurate identification of structural planes. Finally, the position, dip angle, and dip direction can be obtained based on the sine fitting function parameters of the structural plane. The test results show that the proposed method achieves a structural plane recognition rate of 93.7%. This method promotes the transformation of the borehole imaging system from qualitative manual analysis to intelligent quantitative analysis.