Rapid Scene Creation for Multispectral Terrain Signature Models and Simulations

JRM and DVC propose significant innovations around our powerful OSCRE architecture for rapid scene generation directly from raw terrain source data. OSCRE requires: high-resolution DEM data input, complete shape file/vector maps, and high-resolution satellite imagery. With these inputs, OSCRE supports OTF terrain tessellation/creation, 3D feature extrusion and model instantiation, material classification/physical property attribution, and credible physics-based multi-spectral sensor image generation. OSCRE applies physics-based irradiance, reflection, thermal, and atmospherics models to provide accurate, multispectral sensor simulation while rendering in real-time. OSCRE performs this entire process On-The-Fly, from loading the raw source data to producing high-fidelity, high dynamic range multispectral sensor imagery. This SBIR effort expands OSCREs architecture and develops innovative new algorithms to address cases where NGA and other terrain data-sources have gaps, like incomplete, low-resolution, or non-existent vector, feature, and elevation data. Such innovations will include fast photogrammetric algorithms for high-resolution DEM generation from stereo-pair imagery and/or Lidar/Ladar data; advanced, fast algorithms for feature extraction; rapid shape/vector-file generation; and advanced fast color-space clustering and PCA algorithms for image segmentation and material classification. The Phase I design and Phase II prototype will include a mechanism to generate rapid terrain/scene database output compatible with CSG and other multi-spectral sensor image rendering engines.