SynBench
收藏资源简介:
SynBench是由海德堡大学曼海姆医学系曼海姆医学智能系统研究所创建的非刚性三维点云配准合成基准数据集。该数据集包含30个基本对象,涵盖了多种变形级别、噪声、离群值和不完整性挑战,总共有80060个样本。数据集通过SimTool工具生成,提供了变形前后的对应点真值,适用于评估非刚性点云配准方法的鲁棒性。SynBench旨在为未来的非刚性点云配准方法提供一个公平的比较平台,特别是在医学手术和软组织建模等领域。
SynBench is a synthetic benchmark dataset for non-rigid 3D point cloud registration, developed by the Mannheim Institute for Medical Intelligent Systems, Department of Medicine, Heidelberg University. This dataset includes 30 base objects, covering a wide range of challenges such as varying deformation degrees, noise, outliers, and point cloud incompleteness, with a total of 80,060 samples. Generated using the SimTool, the dataset provides ground-truth point correspondences between point clouds before and after deformation, and is designed to evaluate the robustness of non-rigid 3D point cloud registration methods. SynBench aims to establish a fair comparison platform for future non-rigid 3D point cloud registration approaches, particularly in domains like medical surgery and soft tissue modeling.

- 1SynBench: A Synthetic Benchmark for Non-rigid 3D Point Cloud Registration海德堡大学曼海姆医学系曼海姆医学智能系统研究所(MIISM) · 2024年



