Scholarlink

<p>In this paper, based on the improved RMBR (IRMBR) algorithm, we propose Sequential IRMBR (SIRMBR) algorithm to automatically generate more realistic and regular outlines from the non-orthogonal polygon building outlines with jagged edges that automatically extracted from high-resolution satellite images. The proposed SIRMBR extracts line segments that adequately represent the building walls by applying the IRMBR sequentially to the initial jagged building outline while reducing the number of outline points. The extracted line segments are extended, intersected and connected to construct a polygon and we finally obtain the regularized building outline. The experimental results showed that the proposed method can generate regular and realistic building outlines with parallel or perpendicular edges from the jagged building outlines of complex shapes. The area ratio, RMSE, and orientation error values between the SIRMBR-derived building outlines and the initial building outlines were 0.999, 0.19&#xa0;m, and 1.24°, respectively. These values indicate that the proposed SIRMBR can also sufficiently preserve the size, shape and orientation information of the initial jagged building outlines. Additionally, the comparison with the most common and recent methods showed that the SIRMBR is superior in terms of generating more realistic building outlines with balanced shape regularity and data fidelity.</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

Regularization of jagged polygon building outlines using improved RMBR algorithm

  • 期刊论文

  • 发表日期：

    2025年7月25日

查看全文
  • Nam-Hyok Pak,
  • Kum-Chol Om,
  • Ryong-Chol Kang,
  • Yong-Ho Kang,
  • Tae-Song Chon

摘要

In this paper, based on the improved RMBR (IRMBR) algorithm, we propose Sequential IRMBR (SIRMBR) algorithm to automatically generate more realistic and regular outlines from the non-orthogonal polygon building outlines with jagged edges that automatically extracted from high-resolution satellite images. The proposed SIRMBR extracts line segments that adequately represent the building walls by applying the IRMBR sequentially to the initial jagged building outline while reducing the number of outline points. The extracted line segments are extended, intersected and connected to construct a polygon and we finally obtain the regularized building outline. The experimental results showed that the proposed method can generate regular and realistic building outlines with parallel or perpendicular edges from the jagged building outlines of complex shapes. The area ratio, RMSE, and orientation error values between the SIRMBR-derived building outlines and the initial building outlines were 0.999, 0.19 m, and 1.24°, respectively. These values indicate that the proposed SIRMBR can also sufficiently preserve the size, shape and orientation information of the initial jagged building outlines. Additionally, the comparison with the most common and recent methods showed that the SIRMBR is superior in terms of generating more realistic building outlines with balanced shape regularity and data fidelity.