Scholarlink

<p>In elongated underground tunnels, laser scanning systems acquire point clouds containing substantial redundant data, leading to heavy computational load and affecting the accuracy of subsequent data analysis. Therefore, this study proposes a feature-preserving point cloud simplification method based on cylindrical neighborhoods, which leverages parallel processing techniques, drawing on high-performance computing principles, to enable fast data handling and meet tunnel monitoring efficiency requirements. The method downsamples feature-redundant points by removing thickness-redundant points, simplifying the point cloud while preserving structural features. Experimental results show that across simplification rates, over 60% of points have roughness below 0.5&#xa0;cm, and the feature preservation rate exceeds 90%. Compared with the random sampling method (RSM), the uniform simplification method (USM), and the feature-preserving method (FPM), the average roughness can be reduced by 45.8% and effectively improves the feature preservation rate, demonstrating that the proposed method can effectively simplify point clouds while retaining critical structural 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

Feature-preserving point cloud simplification method based on cylindrical neighborhoods

  • 期刊论文

  • 发表日期：

    2025年11月7日

查看全文
  • Yuan Zhang,
  • Shaoyi Xu,
  • Chengtao Wang,
  • Tianshan Sun,
  • Hao Wang

摘要

In elongated underground tunnels, laser scanning systems acquire point clouds containing substantial redundant data, leading to heavy computational load and affecting the accuracy of subsequent data analysis. Therefore, this study proposes a feature-preserving point cloud simplification method based on cylindrical neighborhoods, which leverages parallel processing techniques, drawing on high-performance computing principles, to enable fast data handling and meet tunnel monitoring efficiency requirements. The method downsamples feature-redundant points by removing thickness-redundant points, simplifying the point cloud while preserving structural features. Experimental results show that across simplification rates, over 60% of points have roughness below 0.5 cm, and the feature preservation rate exceeds 90%. Compared with the random sampling method (RSM), the uniform simplification method (USM), and the feature-preserving method (FPM), the average roughness can be reduced by 45.8% and effectively improves the feature preservation rate, demonstrating that the proposed method can effectively simplify point clouds while retaining critical structural features.