Material Classification for Physics-Based Sensor Simulation Using Stereo-Pair Imagery

JRM proposes to leverage its Phase I design work to develop Phase II improvements in its material classification algorithms and software, particularly improvements leveraging spatial data available from stereo pair satellite imaging data sources. Specific Phase II innovations: Improved material classification techniques and software for leveraging 3D stereo-pair-derived data; material-prediction algorithm improvements that help alleviate common color-space material-ambiguities by leveraging spatial-relief discriminators based 3D data like that obtainable from stereo-pair satellite imagery. Super-resolution techniques and algorithms for adding detail beyond source data resolution based on key 3D properties of the material assignment. (i.e. sub-pixel material assignments, surface roughness, DHR variation, elevation maps, etc). Fast, efficient Software I/O techniques for ingesting stereo-pair 3D data, from I/O routines tailored to existing COTS 3D spatial data products (surface normal and DEM) to algorithms and software for processing stereo-pair imagery into 3D-point clouds where refined data does not exist. Improved GPU-techniques for sensor simulation with COTS IG providers for improved EO, IR, and radar channel simulation which leverage the above stereo-pair imagery-derived enhancements. Phase II Simulation and Algorithm Development Testbed, for testing the Phase II improved algorithms against simulated stereo-pair satellite data. Improved material data libraries for producing and exploiting stereo-pair-enhanced material-classified terrain databases.