Strategies for hyperspectral target detection in complex background environments

The capability to rapidly detect and identify ground targets hidden in natural and man-made clutter over broad areas remains of high interest for defense applications, and is emerging as a critical component to counter-terrorism and homeland security missions. Recognizing the difficulty in establishing such a capability based on traditional, panchromatic imaging sensors in the electro-optical and infrared spectral region, the Air Force Research Laboratory (AFRL) has been investigating the added potential that spectral and polarimetric sensing methods provide. This has resulted in a variety of sensor system developments and baseline processing systems for real-time target detection and discrimination. Despite the success of this prior research, however, the emerging missions are expected to present situations for which the targets are smaller, more diverse, and more fleeting, while the background clutter is more complex. This is likely to severely limit the detection performance achievable with the current generation of processing algorithms based on standard statistical background models and require more sophisticated detection and discrimination algorithms. This paper provides an overview of some of the primary AFRL research interests and directions to address these limitations