Scholarlink

<p>Symmetry prior on structure is instrumental for single-view 3D reconstruction. This paper introduces a novel mathematical approach for explicit reconstruction of high-quality 3D models of axially symmetric objects from single images. The proposed algorithm utilizes the symmetry prior to solve an optimization problem for finding the optimal 3D axis of rotational symmetry such that the projection of the reconstructed 3D body matches the 2D input image closely. Additionally, we introduce a method for texture acquisition from a single image of the surface of revolution, aiming to create more realistic and detailed 3D models suitable for applications in computer animation and virtual reality. Our results, both qualitatively and quantitatively, demonstrate the effectiveness of our framework in reconstructing precise textured 3D models. The code is available at <a href="https://smh-hosseiny.github.io/SOR-3D/">https://smh-hosseiny.github.io/SOR-3D/</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

Single-view 3D reconstruction of surface of revolution

  • 期刊论文

  • 发表日期：

    2025年4月28日

查看全文
  • Seyed M. Hosseini,
  • Seyed-Mahdi Nasiri,
  • Reshad Hosseini,
  • Hadi Moradi

摘要

Symmetry prior on structure is instrumental for single-view 3D reconstruction. This paper introduces a novel mathematical approach for explicit reconstruction of high-quality 3D models of axially symmetric objects from single images. The proposed algorithm utilizes the symmetry prior to solve an optimization problem for finding the optimal 3D axis of rotational symmetry such that the projection of the reconstructed 3D body matches the 2D input image closely. Additionally, we introduce a method for texture acquisition from a single image of the surface of revolution, aiming to create more realistic and detailed 3D models suitable for applications in computer animation and virtual reality. Our results, both qualitatively and quantitatively, demonstrate the effectiveness of our framework in reconstructing precise textured 3D models. The code is available at https://smh-hosseiny.github.io/SOR-3D/.