Automatic Construction of Virtual Worlds

There is increasing interest in developing virtual world databases for advanced distributed simulation that maintain a high degree of fidelity with respect to the actual three dimensional environment. The compilation of such spatial databases requires the integration of information from a variety of sources, including digital map data, aerial imagery, detailed line drawings, and ground based photography. Current applications for such databases have revolved around distributed interactive simulation addressing military training, primarily in the areas of mission planning and rehearsal and in vehicle simulation.

Factors that currently limit the development of such systems include the lack of terrain and environmental data at the level of detail required for ground-based simulations, the magnitude of the scene generation process, and the inadequacy of input devices that allow a user to interact with the virtual world.

Much of the ongoing research on automated feature extraction at the MAPSLab can be applied to the construction or intensification of simulation databases. For example, manual compilation of building models and road networks is particularly time-consuming, and automated and semi-automated analysis methods can speed up the database intensification process. Our research in multispectral analysis is particularly appropriate for increasing simulation fidelity, in its ability to establish surface material categories for improved visualization of man-made and natural features.

We have been primarily concerned with the generation of virtual world databases, and have not investigated real-time user interaction. The issues we have addressed are:

• The generation of an efficient terrain model, a triangular irregular network, (TIN) suitable for real-time visualization.

• The integration of road networks with the TIN model to form a coherent terrain skin. Integrating features such as roads, rivers, lakes, and forests into the terrain skin results in a coherent spatial model that can be used to guarantee flat water lakes and trafficable roads even in areas of high relief.

• The control of polygon complexity in urban areas that might otherwise overload simulation hardware, taking both road and terrain complexity into account.

• The integration of cultural features extracted from aerial imagery with existing spatial data, possibly at extreme differences in spatial resolution. Spatial database intensification requires a rigorous geodetic model and the ability to perform traditional cartographic tasks such as feature generalization, aggregration, and replacement.