Search

In addition to the traditional Geographic Information System (GIS) data such as images and vectors, point cloud data has become more available. It is appreciated for its precision and true three-Dimensional (3D) nature. However, managing the point cloud can be difficult due to scaling problems and specificities of this data type. Several methods exist but are usually fairly specialised and solve only one aspect of the management problem. In this work, we propose a complete and efficient point cloud management system based on a database server that works on groups of points rather than individual points. This system is specifically designed to solve all the needs of point cloud users: fast loading, compressed storage, powerful filtering, easy data access and exporting, and integrated processing. Moreover, the system fully integrates metadata (like sensor position) and can conjointly use point clouds with images, vectors, and other point clouds. The system also offers in-base processing for easy prototyping and parallel processing and can scale well. Lastly, the system is built on open source technologies; therefore it can be easily extended and customised. We test the system will several billion points of point clouds from Lidar (aerial and terrestrial ) and stereo-vision. We demonstrate ~ 400 million pts/h loading speed, user-transparent greater than 2 to 4:1 compression ratio, filtering in the approximately 50 ms range, and output of about a million pts/s, along with classical processing, such as object detection.

Streets are large, diverse, and used for conflicting transport modalities as well as social and cultural activities. Proper planning is essential and requires data. Manually fabricating data that represent streets (street reconstruction) is error-prone and time consuming. Automatising street reconstruction is a challenge because of the diversity, size, and scale of the details (~ cm for cornerstone) required. The state-of-the-art focuses on roads and is strongly oriented by each application (simulation, visualisation, planning). We propose a unified framework that works on real Geographic Information System (GIS) data and uses a strong, yet simple hypothesis when possible to produce coherent street modelling at the city scale or street scale. Because it is updated only locally in subsequent computing, the result can be improved by adapting input data and the parameters of the model. We reconstruct the entire Paris streets in a few minutes and show how the results can be edited simultaneously by several concurrent users.