GMTI Radar Target Classification Using Spectral and Tracking Data

TSC will develop Human, Vehicle, Animal and Clutter (HVAC) discrimination algorithms for use in GMTI radars by leveraging research performed on previous DARPA efforts and an ongoing Army program. TSC will focus on the challenging UHF radar problem and develop multiple-CPI processing algorithms in Phase I for application to Army airborne foliage penetration surveillance systems and ground-based UHF radar towers that are under development. TSC will utilize the UHF radar database that it collected on humans and diverse animal species for testing as well as other available radar data with the Armys permission. TSC will determine viable Doppler spectral features for reliable HVAC classification and investigate variants of Support Vector Machines and temporal and statistical processing approaches in the base effort. TSC will also develop methods of estimating the classification accuracy of the HVAC discrimination decision. Algorithm performance will be evaluated for different targets, foliage environments, wind conditions and other factors. In the option effort, TSC will investigate the use of long-term track features for HVAC discrimination and develop methods to recognize multiple targes. In Phase II, TSC will mature these algorithms, expand them to other radar bands, and validate them on collected airborne and tower radar data as available.