Using Machine Vision Techniques to Create a Multi-Resolution Obstacle Database

The Navy requires an architecture for acquiring and storing 3-Dimensional obstacle data. An extension of the Terrain Awareness Warning System (eTAWS), the obstacle database must store man-made objects such as buildings, bridges, towers, overpasses, and wires. The system must be able to generate obstacle database data from a number of sources including synthetic aperture radar (SAR), stereo/multiple aerial/satellite photography, and raw geodetic survey data.Cybernet proposes to focus on the collection of 3D obstacle geometry/location based on aerial photography and using advance image processing techniques. This collection process will then be used to populate an indexed database that stores obstacle data at multiple resolutions. By indexing the data, we will be able to reduce the storage requirements by storing redundant data only once. By storing the data at multiple resolutions we allow the application accessing it to balance the tradeoffs corresponding to the different levels of resolution/detail, such as resolution vs. coverage area.