Automatic Extrusion of Surface Features from Terrain Aerial/Satellite Imagery

The purpose of this SBIR topic is to explore and implement technologies capable of automatically extracting three dimensional feature sets from aerial imagery. Successful implementation of these technologies will speed the process and decrease the cost of developing complex, geospatially accurate databases for military training and mission rehearsal. In Phase I of this SBIR effort, SimWright successfully demonstrated the ability to automatically identify and extract geospatially referenced three-dimensional cultural features. The capability of these algorithms was demonstrated using both aerial and satellite imagery and the process is compatible with a broad range of both commercial and national means imagery formats. Phase II implementation requires creation of functions that can be integrated into commercial applications used for terrain database generation. Research will be conducted to develop applications with the following functionality: 1)An automated feature modeler; 2)An automated large area building extractor;3)An automated single building extractor; 4)An automated feature attribution; 5)Enhancements to an existing Complex Building; Generator; and 6) Development of a feature model/database previewer for use in an existing terrain database application Phase I focused on spatial feature extraction parameters. In Phase II there will be emphasis on spectral feature parameters for automated feature attribution.