Free and Open Source Software for Geomatics Conference FOSS4G 2010 Barcelona

Selected Presentations

Home > Presentations > Abstract details




Polygonal meshes are powerful structures to represent geometric information of the Earth’s surface. In particular, triangle meshes have been massively used as a reliable way to efficiently represent the land surface with real time responses in virtual navigation. In this work we present new ideas for the underlying treatment of a mesh that improve efficiency and quality in the navigation. We explore critical issues such as on-the-fly refinement, smoothness of the mesh and partition methods for the refinement operations. We test the proposed capabilities inside an open source virtual globe and compare with other implementations. The proposed model is a view-dependent level of detail strategy that is classified as a binary triangle tree approach. We study how to increase the accuracy of the mesh by splitting the triangles by their longest edge and propagating the splits to the surrounding neighbours. This binary triangle tree naturally avoids the cracks on the terrain. There is also an inverse merging process that simplifies the mesh at the same time that it maintains a high visual quality. This approach allows a high frame rate and the frame-to-frame coherence is maintained thanks to the split and merge policy. In this application, the textures and meshes of the terrain are obtained progressively from a remote server using standard protocols. This information is conveniently treated and mapped into an ellipsoidal model representing the Earth’s geoid. During the user navigation, the application may manage huge amounts of information that are served from the remote server in regular tiles. The information of these tiles is extracted to refine or simplify the mesh. The land textures of the tiles are organized in a quad tree data structure at different levels of resolution. The correspondence between the vertices of the mesh and the texture coordinates are established in run time.



Javier Sánchez - University of Las Palmas de Gran Canaria
Jóse Pablo Suárez - University of Las Palmas de Gran Canaria
Agustín Trujillo - University of Las Palmas de Gran Canaria
Modesto Castrillón - University of Las Palmas de Gran Canaria