Automatic Terrain Characterization and Feature Identification in FOPEN SAR Imagery

The new FOPEN SAR ATD testbed and future Global Hawk UAV-based operational system being developed by DoD will have the ability to detect targets concealed beneath the foliage. Area delimitation and local context exploitation are critical elements of thisnew surveillance capability. TSC proposes to refine and extend the innovative algorithms that were developed in Phase I to estimate topography, classify terrain cover including roads, and characterize the foliage. Our approach for DEM generation is toemploy stereo processing that takes advantage of multiple looks, existing DTED and strong tree trunk returns to achieve high accuracy. Collected data from the ATD testbed will be employed for algorithm training and evaluation. TSC will utilize theextracted contextual information as input to mobility metrics that can reduce the search area and decrease the number of false alarms in forested areas. TSC will also investigate the use of auxiliary imagery as well as other data sources, and leverage ourexisting stereo DEM generation, Automatic Terrain Classification and mobility assessment tools developed on related efforts to achieve the program objectives. At the conclusion of the effort, TSC will deliver a comprehensive PC software toolkit thatincorporates all of the new algorithm capabilities, and can be used to process collected ATD data.