The One World Terrain (OWT) project seeks to create a digital twin of the entire world for use in US Army simulators. As part of an ongoing research effort within DEVCOM Soldier Center Simulation & Training Technology Center, High-Resolution Insets are being added to the overall terrain inventory using photogrammetry-based methods with drone-captured photos and LiDAR data. To produce high-resolution inset data, a pipeline of processing stages takes place after the initial creation of terrain meshes and textures from the possibly thousands of drone-captured photos of the area. A crucial stage of the pipeline involves identification, extraction, and optimization of building exteriors for subsequent use. Currently, the building-processing stage creates reference material for artists to manually create accurate 3D models of the buildings. Simultaneously, efforts are underway to fully automate the building-processing stage. This paper will acquaint the reader with the unique output of the photogrammetry process, highlighting the large gap between the raw output and what is needed for today’s simulators. The paper then focuses on several novel techniques within the building-processing stage that eliminate noise and reduce complexity. These techniques support the growing and evolving demands for higher-resolution modeling and simulation for Live training.

Keywords
AUTOMATION;CONTENT GENERATION;TERRAIN

Additional Keywords
Photogrammetry