Advanced ATC for Tactical Information Processing

The proposed effort is directed at achieving Automatic Target Cueing (ATC) performance at levels which are essential for near term system deployment. The performance required for deployment is beyond the capability of current systems. The design emphasis is to have robust performance under a full range of operating conditions, and to use techniques which lead to efficient real time implementation in Phase II. The basis of the approach is to apply adaptive nonlinear spatial filtering to FLIR imagery. This innovative filtering allows the imagery to be simultaneously characterized in terms of both local thermal intensity and target spatial extent. We have shown static (non-adaptive) filters of this type can detect very low contrast targets in real imagery with sensitivity superior to other architectures. The proposed approach uses approximate target range information to allow real time adaptation of the filter parameters. This innovation results in sharper filtering, such that the filter response to typical clutter is suppressed, while a strong response to targets is retained. The overall result will be a significant reduction in ATC system false alarm rates, while maintaining the capability to detect low signature targets. In the phase I effort, we will adapt proven range independent FLIR ATC algorithms to utilize data from a laser rangefinder. This system will be tested and evaluated on selected FLIR imagery. Additionally, textural and structural metrics will be developed which can be used to reject false alarms resulting from thermal gradients caused by natural objects. In the Phase II program, the full ATC system will be refined, and implemented in real time hardware. Performance will be evaluated in a blind test on a comprehensive set of imagery which encompasses a wide variety of targets, target ranges, and terrain types. Anticipated benefits of the proposed research include dramatically improved performance for ground based and airborne sensor platforms for detection, cueing, and target servicing tasks. This research can be used to aid in the evaluation of the quality and characteristics of simulated imagery. Other applications which will benefit include low-flying aircraft obstacl