Scholarlink

<p>Instant access to the particle size characteristics of sandstone is of great importance to improve the efficiency and safety of deep in situ mining. It is highly challenging to quantify the particle size under non-destructive conditions. In this research, six distance-dependent texture features were further extracted via the gray-scale level co-occurrence matrix (GLCM) analysis. The gray-scale distribution reveals that the extremely reflective pixels are prevalent in the gray-scale images of the sandstone, accounting for an average of 0.06% of each image. The image preprocessing can effectively enhance the dynamic range, the global contrast, and the most dominant gray-scale values. Depending on the variation pattern with the distance, all six features were divided into scale-sensitive features and scale-invariant features. A prediction model for the median particle size of sandstone based on the texture feature scale was finally developed and validated. The results indicate entropy exhibits the best prediction performance with a maximum relative error of 15.8%. These achievements could provide a theoretical basis and an effective solution for predicting the physical and mechanical properties of the sandstone by texture features.</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

Multi-distance gray-scale level co-occurrence matrix characteristics and particle size prediction model of sandstone

  • 期刊论文

  • 发表日期：

    2026年2月1日

查看全文
  • Rui Xu,
  • Xiaolong Ma,
  • Yiteng Wang,
  • Yong Wan,
  • Ming Bai,
  • Houzhen Wei

摘要

Instant access to the particle size characteristics of sandstone is of great importance to improve the efficiency and safety of deep in situ mining. It is highly challenging to quantify the particle size under non-destructive conditions. In this research, six distance-dependent texture features were further extracted via the gray-scale level co-occurrence matrix (GLCM) analysis. The gray-scale distribution reveals that the extremely reflective pixels are prevalent in the gray-scale images of the sandstone, accounting for an average of 0.06% of each image. The image preprocessing can effectively enhance the dynamic range, the global contrast, and the most dominant gray-scale values. Depending on the variation pattern with the distance, all six features were divided into scale-sensitive features and scale-invariant features. A prediction model for the median particle size of sandstone based on the texture feature scale was finally developed and validated. The results indicate entropy exhibits the best prediction performance with a maximum relative error of 15.8%. These achievements could provide a theoretical basis and an effective solution for predicting the physical and mechanical properties of the sandstone by texture features.