ScanVan Project : Images and 3D cloud points

The ScanVan project (2017-2021) aimed at demonstrating that Spherical Camera can greatly simplify the production of 3D photogrammetric datasets, allowing computing 3D models of cities on a daily basis. The initial phase of the project was conducted by two complementary teams, one at EPFL and one at HES-SO Valais. The EPFL team worked on the theoretical aspects and mathematical processing of algorithms that compute 3D scenes from spherical images and on the design the iterative process for doing the computation of large-scale graphs of connected scenes to process 3D models of an entire city. The HES-SO Valais team worked on the design and physical implementation of the system. In the course of the project, the research teams designed an omnidirectional central camera, an algorithm for geometry estimation that operates directly on spherical images and a densification technique based on the estimated geometry. The camera was mounted a vehicle and tested in the city of Sion. The dataset contains : - 4 models of the same loop of Sion : - a sparse model in ply format - a dense model in ply format - the sparse geo-registered model in the swiss coordinate system (EPSG:2056, CH1903+) - the dense geo-registered model in the swiss coordinate system (EPSG:2056, CH1903+) - spherical images used for constructing this model, acquired with the ScanVan central omnidirectional camera. For privacy reasons, people and vehicles have been blurred. - the *geodesy* folder contains - a csv file listing the latitude and longitude of all images as positionned in the geo-registered model - a text file listing the homologous points which were used to geo-register the model from its own coordinate system to the CH1903+ coordinates system - the script which produced the sparse and dense models and the output logs. - various outputs of the 3D sparse computation can be found in the *omvg* folder.ScanVan项目（2017-2021年）旨在验证球形相机（spherical camera）可大幅简化三维摄影测量数据集的生产流程，使城市三维模型的日常规模化生成成为可能。本项目初期由两个互补团队协同开展工作：其一为瑞士联邦理工学院洛桑分校（EPFL）团队，其二为瑞士瓦莱高等专业学院（HES-SO Valais）团队。EPFL团队负责从球形图像计算三维场景的算法理论研究与数学处理，并设计了迭代流程以构建大规模连通场景图，进而完成整座城市三维模型的处理任务。HES-SO Valais团队则负责系统的整体设计与物理实现。 项目研发期间，研究团队研发了一款全向中心相机（omnidirectional central camera）、一套可直接对球形图像运行的几何估计算法，以及一种基于估计几何的稠密化技术。该相机被搭载于车辆之上，并在锡永市（Sion）完成了实地测试。 本数据集包含以下内容： - 4套锡永同一路段的三维模型： - PLY格式的稀疏模型 - PLY格式的稠密模型 - 采用瑞士坐标系（EPSG:2056, CH1903+）的地理配准稀疏模型 - 采用瑞士坐标系（EPSG:2056, CH1903+）的地理配准稠密模型 - 用于构建上述模型的球形图像，由ScanVan中心全向相机采集，出于隐私保护需求，所有图像中的人物与车辆均已做模糊处理。 - `geodesy`文件夹包含： - 一份CSV文件，列出了地理配准模型中所有图像的经纬度坐标 - 一份文本文件，列出了用于将模型从自身坐标系转换至CH1903+坐标系的同名点 - 用于生成稀疏与稠密模型的脚本，以及对应的输出日志 - `omvg`文件夹中存储了三维稀疏计算的各类输出结果。