For storing and modelling three-dimensional topographic objects (e.g. buildings, roads, dykes and the terrain), tetrahedralisations have been proposed as an alternative to boundary representations. While in theory they have several advantages, current implementations are either not space efficient or do not store topological relationships (which makes spatial analysis and updating slow, or require the use of a costly 3D spatial index). We discuss in this paper an alternative data structure for storing tetrahedralisations in a DBMS. It is based on the idea of storing only the vertices and stars of edges; triangles and tetrahedra are represented implicitly. It has been used previously in main memory, but not in a DBMS - we describe how to modify it to obtain an efficient implementation in a DBMS. As we demonstrate with one real-world example, the structure is around 20 % compacter than implemented alternatives, it permits us to store attributes for any primitives, and has the added benefit of being topological. The structure can be easily implemented in most DBMS (we describe our implementation in PostgreSQL) and we present some of the engineering choices we made for the implementation. © 2013 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Ledoux, H., & Meijers, M. (2013). Representing three-dimensional topography in a DBMS with a star-based data structure. In Lecture Notes in Geoinformation and Cartography (pp. 119–132). Kluwer Academic Publishers. https://doi.org/10.1007/978-3-642-29793-9_7
Mendeley helps you to discover research relevant for your work.