Scholarlink

<p>3D point cloud registration is important in robotics and computer vision to find a&#xa0;rigid geometrical transformation to align a&#xa0;pair of point clouds with unknown point correspondences. There are many applications of point cloud registration in localization, scene reconstruction, scene flow estimation, and autonomous driving. The most common traditional registration method is the iterative closest point (ICP), which alternates between two steps: solving point correspondences and orthogonal transformation. The drawback of ICP is that it is sensitive to initialization and often converges to incorrect local minima. Some modified ICP variants can still easily fall into local minima in case of noisy or partially overlapping point clouds. In the 3D scene reconstruction task, the sequence of the pairwise registrations is considered. In this case, translation estimation errors often make an essential contribution to the overall error. In this paper, we propose an algorithm to align the two point clouds based on a&#xa0;new multiparameter functional for the estimation of rotations and translations, a&#xa0;closed-form algorithm is used. Computer simulation uses real and synthetic data provided to illustrate the performance of the proposed method.</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

A multiparameter functional for rotation and translation estimation in point cloud registration task

  • 期刊论文

  • 发表日期：

    2025年12月15日

查看全文
  • Artyom Makovetskii,
  • Sergei Voronin,
  • Vitaly Kober,
  • Alexei Voronin

摘要

3D point cloud registration is important in robotics and computer vision to find a rigid geometrical transformation to align a pair of point clouds with unknown point correspondences. There are many applications of point cloud registration in localization, scene reconstruction, scene flow estimation, and autonomous driving. The most common traditional registration method is the iterative closest point (ICP), which alternates between two steps: solving point correspondences and orthogonal transformation. The drawback of ICP is that it is sensitive to initialization and often converges to incorrect local minima. Some modified ICP variants can still easily fall into local minima in case of noisy or partially overlapping point clouds. In the 3D scene reconstruction task, the sequence of the pairwise registrations is considered. In this case, translation estimation errors often make an essential contribution to the overall error. In this paper, we propose an algorithm to align the two point clouds based on a new multiparameter functional for the estimation of rotations and translations, a closed-form algorithm is used. Computer simulation uses real and synthetic data provided to illustrate the performance of the proposed method.