Efficient Visibility Determination in Urban Scenes Considering Terrain Information

Abstract

In this article, we introduce a novel occlusion culling method working on the server side to provide real-time navigation on web-based systems. Nowadays, virtual navigation in urban environments is a rising trend in several contexts such as tourism, GPS navigation systems, and video games. A city environment is usually associated with a complex data model that is better stored, maintained, and updated on a server system. Mobile devices are regular clients in these cases, demanding this information in a fast, reliable, and engaging way. Even though these gadgets have been increasing their capabilities in computation and visualization, the bottleneck is still the transmission of information over the network. The advantage of urban environments is that, from a user viewpoint, only a small portion of the scene is visible. This feature makes crucial the use of occlusion culling techniques working on the server side in order to transmit to the client side only the small set of visible elements compared to the whole scene. The input data are the city geometry from the 2D cadastral information system, the building textures, and DEM (Digital Elevation Model) files with the urban terrain features. In a first stage, the process creates a 2.5D urban model with all these data in preprocessing time. Then the client provides the user location point, and the server sends back the exact portion of visible city. This approach is implemented using polar diagrams for visibility determination and LOD (Level of Detail) techniques for further geometry reduction.