Imagery Manipulation for Simulator Databases

The goal for Phase II is to develop tools that provide significant improvements in automated database generation from imagery. These tools will focus on the automatic removal of time-specific artifacts including shadows and vehicles. They will also beused to extract information from the image to generate and insert correlated 3D cultural models such as building, trees, and other objects into the generated terrain database. These tools will fill an important gap in current technology because they willimprove the database fidelity and reduce the associated time and costs. Current tools require significant manual intervention to perform the proposed tasks.Diamond Visionics successfully automated shadow identification and anticipates automated shadow removal within aerial imagery in Phase I. We explored methods of time-specific object detection, removal, and subsequent object insertion methods. In Phase IIwe will develop the tools and integrate with existing COTS products, allowing high fidelity simulator databases to be quickly created from overhead imagery. The effort will include a technology demonstration at the end of Phase II. This demonstrationwill start with representative high altitude imagery and follow the highly automated process of creating a visual database that incorporates image manipulation as well as 3D model generation.