LidarScout
收藏资源简介:
LidarScout 是一种用于探索大规模压缩点云数据集的方法,能够实时可视化数百亿个点的国家规模扫描。数据集包括来自美国、新西兰和瑞士的点云数据,大小从 1.6 亿个点到 2620 亿个点不等。数据集以压缩的 LAZ 格式存储,无需预处理即可快速加载稀疏子样本,并生成高质高程图。该方法主要用于解决大规模点云数据的实时可视化问题。
LidarScout is a method for exploring large-scale compressed point cloud datasets, enabling real-time visualization of nation-scale scans containing tens of billions of points. The dataset encompasses point cloud data collected from the United States, New Zealand, and Switzerland, with scale ranging from 160 million points to 262 billion points. Stored in the compressed LAZ format, the datasets support rapid loading of sparse subsamples without preprocessing, and enable the generation of high-quality elevation maps. This method is primarily developed to tackle the real-time visualization challenge of large-scale point cloud data.
LidarScout 数据集概述
数据集基本信息
- 名称:LidarScout
- 官方实现:High-Performance Graphics 2025论文的官方实现
- 主要功能:直接外存渲染大规模点云
数据集内容与用途
- 数据格式:支持LAZ格式的点云数据
- 数据规模:可处理TB级别的压缩点云数据
- 技术特点:
- 通过快速加载边界框实现大规模分块数据集的加载
- 仅加载屏幕上显示较大的边界框内的点
- 动态加载和卸载点云数据
- 通过点云稀疏子样本预测高度图显示闭合表面
数据来源
论文中使用的点云数据集来源:
- CA13_SAN_SIM:https://portal.opentopography.org/datasetMetadata?otCollectionID=OT.032013.26910.2
- Bund_BoraPk:https://portal.opentopography.org/datasetMetadata?otCollectionID=OT.052019.6341.1
- ID15_Bunds:https://portal.opentopography.org/datasetMetadata?otCollectionID=OT.112020.6341.1
- NZ23_Gisborne:https://portal.opentopography.org/datasetMetadata?otCollectionID=OT.022024.2193.1
- BR17_SaoPaulo:https://portal.opentopography.org/datasetMetadata?otCollectionID=OT.062020.31983.1
- swissSURFACE3D:https://www.swisstopo.admin.ch/de/hoehenmodell-swisssurface3d
相关资源
- 论文地址:https://www.cg.tuwien.ac.at/research/publications/2025/erler-2025-lidarscout/
- 训练代码仓库:https://github.com/cg-tuwien/lidarscout_training
- Windows可执行文件:https://users.cg.tuwien.ac.at/perler/lidarscout/LidarScout.zip
引用格式
bibtex @inproceedings{erler2025lidarscout, booktitle = {High-Performance Graphics - Symposium Papers}, editor = {Knoll, Aaron and Peters, Christoph}, title = {{LidarScout: Direct Out-of-Core Rendering of Massive Point Clouds}}, author = {Erler, Philipp and Herzberger, Lukas and Wimmer, Michael and Schütz, Markus}, year = {2025}, publisher = {The Eurographics Association}, ISSN = {2079-8687}, ISBN = {978-3-03868-291-2}, DOI = {10.2312/hpg.20251170} }

- 1LidarScout: Direct Out-of-Core Rendering of Massive Point Clouds维也纳工业大学 · 2025年



