Scholarlink

<p>In the recent years, there have been numerous attempts to quantify the porosity of sandstones by different computational algorithms, and to a lesser extent to identify minerals in petrographic thin sections using diverse techniques. This is a challenging task due to the optical similarities between framework grains depending on the relative position of their birefringence axis in relation to the axis of the lower (illumination) and upper (detection) polarizers of the microscope. In this work, we present a new workflow consisting in the acquisition of one plane-polarized light image and seven cross-polarized light images, to increase the optical information present in the digital analysis of siliciclastic-dominated sandstones and then generate a new synthetic Principal Component Analysis image. This technique was implemented in lithic feldsarenites and feldspathic litharenites from the Jurassic Lajas Formation (a hydrocarbon reservoir unit of the Neuquén Basin, Argentina), enabling a fully automated segmentation with a panoptic quality of 0.61 using this workflow, compared to 0.36 done using a plane polarized light image. Additionally, a fast (few minutes) manual post-processing increases the panoptic quality to 0.97. This shows that the use of more than one or two images increases the variability of the grain boundaries, allowing a more complete segmentation of the components that make up the solid matrix structure. Based on the segmentation results, several parameters, including petrographic content, as well as grain size, distribution, and contacts, were determined and statistically analyzed. This segmentation will be the starting point for a more detailed granulometric, textural, and petrographic classification of the rock as well as the resolution of its associated porosity.</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

Automated segmentation using computer vision applied to petrographic thin sections of unconventional tight gas sandstone reservoirs

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月10日

查看全文
  • Lucrecia Frayssinet,
  • Patricio Grinberg,
  • Hernán E. Grecco,
  • Marcos A. Comerio,
  • Esteban A. Domené

摘要

In the recent years, there have been numerous attempts to quantify the porosity of sandstones by different computational algorithms, and to a lesser extent to identify minerals in petrographic thin sections using diverse techniques. This is a challenging task due to the optical similarities between framework grains depending on the relative position of their birefringence axis in relation to the axis of the lower (illumination) and upper (detection) polarizers of the microscope. In this work, we present a new workflow consisting in the acquisition of one plane-polarized light image and seven cross-polarized light images, to increase the optical information present in the digital analysis of siliciclastic-dominated sandstones and then generate a new synthetic Principal Component Analysis image. This technique was implemented in lithic feldsarenites and feldspathic litharenites from the Jurassic Lajas Formation (a hydrocarbon reservoir unit of the Neuquén Basin, Argentina), enabling a fully automated segmentation with a panoptic quality of 0.61 using this workflow, compared to 0.36 done using a plane polarized light image. Additionally, a fast (few minutes) manual post-processing increases the panoptic quality to 0.97. This shows that the use of more than one or two images increases the variability of the grain boundaries, allowing a more complete segmentation of the components that make up the solid matrix structure. Based on the segmentation results, several parameters, including petrographic content, as well as grain size, distribution, and contacts, were determined and statistically analyzed. This segmentation will be the starting point for a more detailed granulometric, textural, and petrographic classification of the rock as well as the resolution of its associated porosity.