Robust Spatial-Temporal Clutter Rejection Filters and Track-Before-Detect Architectures for Space Based Platforms

Efficient clutter rejection is a challenge for both space-based geostationary and low-earth orbit infrared sensors that must find, extract, and track moving targets in heavily cluttered backgrounds. In such systems, reliable target tracking using conventional approaches fails without clutter rejection down to/below the sensor noise level. The task is complicated by sensor motion which causes image translation, rotation, and non-stationarity that lead in turn to an increased false alarm rate. To address this challenge we propose to develop: (a) Robust, nonlinear spatial-temporal image processing techniques for clutter rejection and image stabilization that are efficient for both geostationary and low-earth orbit platforms and that can be easily integrated into an adaptive multi-filter bank architecture; (b) Extremely accurate two-stage super-stabilization algorithms; (c) Optimal nonlinear filtering-based track-before-detect architectures to detect very dim targets; and (d) Change detection-based advanced track confirmation/deletion algorithms. We will demonstrate that the proposed nonlinear spatial-temporal methods are particularly efficient for target detection/tracking in difficult scenarios where conventional spatial filters and differencing techniques fail. The developed software tools will be capable of predicting the performance and reducing to a minimum the time needed to search for an adequate system configuration in MDA programs/systems.