Scholarlink

<p>Mesh deformation plays a critical role in various engineering simulations applications. Traditional mesh deformation techniques often face challenges such as element inversion, low-quality mesh elements, and distortion during significant geometric boundary changes. This paper proposes a novel semi-supervised learning-based approach for 2D quadrilateral mesh deformation that allows for the tangential sliding of mesh nodes along boundary regions while preserving the topological structure of the boundary. The method leverages deep learning techniques to compute displacements of mesh nodes through specialized loss functions. Through validation with various deformation scenarios, it is demonstrated that the approach shows significant improvements in the robustness and adaptability of mesh deformations, laying the groundwork for future research in deep learning applications for mesh processing.</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

Deep learning-based semi-supervised approach for high-quality 2D quadrilateral mesh deformation

  • 期刊论文

  • 发表日期：

    2025年4月27日

查看全文
  • Shanshan Li,
  • Qian Li,
  • Kaiting Li,
  • Liang Sun

摘要

Mesh deformation plays a critical role in various engineering simulations applications. Traditional mesh deformation techniques often face challenges such as element inversion, low-quality mesh elements, and distortion during significant geometric boundary changes. This paper proposes a novel semi-supervised learning-based approach for 2D quadrilateral mesh deformation that allows for the tangential sliding of mesh nodes along boundary regions while preserving the topological structure of the boundary. The method leverages deep learning techniques to compute displacements of mesh nodes through specialized loss functions. Through validation with various deformation scenarios, it is demonstrated that the approach shows significant improvements in the robustness and adaptability of mesh deformations, laying the groundwork for future research in deep learning applications for mesh processing.