Scholarlink

<p>Multi-view RGBD 3D reconstruction of dynamic scenes presents significant challenges in computer vision. Traditional non-rigid reconstruction methods struggle with large topological changes and rely on memory-intensive volumetric structures. This paper proposes a surfel-based multi-view dynamic reconstruction approach with cross-view smoothing. Our method maintains only surface elements (surfels) for multi-view fusion, reducing computational requirements while enhancing reconstruction quality in cross-angular regions. Experimental results demonstrate that our algorithm effectively handles complex topological changes and smooths cross-view junctions, providing a robust solution for real-time dynamic reconstruction of non-rigid scenes. Code and dataset are available at <a href="https://github.com/RKGalaxy-Luo/MultiViewDynRecon">https://github.com/RKGalaxy-Luo/MultiViewDynRecon</a>.</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-view dynamic reconstruction with cross-view smoothing based on surfel

  • 期刊论文

  • 发表日期：

    2025年12月18日

查看全文
  • Shiguang Han,
  • Xinhua Liu,
  • Zehua Dong,
  • Jiaxuan Luo,
  • Xiaolin Ma,
  • Hailan Kuang

摘要

Multi-view RGBD 3D reconstruction of dynamic scenes presents significant challenges in computer vision. Traditional non-rigid reconstruction methods struggle with large topological changes and rely on memory-intensive volumetric structures. This paper proposes a surfel-based multi-view dynamic reconstruction approach with cross-view smoothing. Our method maintains only surface elements (surfels) for multi-view fusion, reducing computational requirements while enhancing reconstruction quality in cross-angular regions. Experimental results demonstrate that our algorithm effectively handles complex topological changes and smooths cross-view junctions, providing a robust solution for real-time dynamic reconstruction of non-rigid scenes. Code and dataset are available at https://github.com/RKGalaxy-Luo/MultiViewDynRecon.