Scholarlink

<p>The Quadric Error Metrics (QEM) algorithm is widely used for 3D mesh simplification, but it struggles with models containing discontinuous appearance attributes and often fails to preserve fine local details. To address these issues, we propose a textured mesh simplification method that incorporates multiple feature constraints. The approach introduces tailored edge collapse rules and a new error metric, the seam angle error, to handle discontinuous texture regions. In addition, vertex sharpness and texture complexity are employed to enhance the algorithm’s sensitivity to geometric and textural details. Experimental results demonstrate that the proposed method effectively reduces texture distortion at seams. Specifically, at an 80% simplification rate, our method reduces the texture root mean square (RMS) error by approximately 78% compared to T-QEM. While this approach incurs a slight increase in global geometric error, it significantly preserves local visual details and texture continuity.</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

Detail-preserving simplification of textured mesh models for natural objects

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月16日

查看全文
  • Liang Bo,
  • Yuangang Liu,
  • Li Shaohua,
  • Ao Jingcheng

摘要

The Quadric Error Metrics (QEM) algorithm is widely used for 3D mesh simplification, but it struggles with models containing discontinuous appearance attributes and often fails to preserve fine local details. To address these issues, we propose a textured mesh simplification method that incorporates multiple feature constraints. The approach introduces tailored edge collapse rules and a new error metric, the seam angle error, to handle discontinuous texture regions. In addition, vertex sharpness and texture complexity are employed to enhance the algorithm’s sensitivity to geometric and textural details. Experimental results demonstrate that the proposed method effectively reduces texture distortion at seams. Specifically, at an 80% simplification rate, our method reduces the texture root mean square (RMS) error by approximately 78% compared to T-QEM. While this approach incurs a slight increase in global geometric error, it significantly preserves local visual details and texture continuity.