Tracklet Fusion for strategic GMTI

"Current methods of tracking high-value, ground-moving targets often fail to provide continuous tracks due to extended periods of missed detections that result from terrain blockage, foliage attenuation, or target line-of-sight velocities that fall belowthe sensor minimum detectable velocity. When this occurs, the tracks are eventually lost and terminated thus forming short duration track files or tracklets. Terminated tracks adversely affect situational awareness since information that was accumulated oncritical targets is lost. Even if automatic track initialization occurs after the target is redetected, no techniques are currently available to provide high-confidence association of the new track with the old tracklet to reestablish the information.TSC proposes to develop innovative techniques that exploit all available information to reliably link all tracklets within an archived database. Methods will also be implemented to fuse the relevant information and reestablish situational awareness. TSCwill extend a dynamic programming target tracking technique that exploits contextual information and target features to provide robust linkage of track segments (tracklets) for evidence accumulation. During Phase I, TSC will develop tracklet fusiontechniques and evaluate their performance using simulated GMTI data. During Phase II, the techniques will be refined, demonstrated using ground-truth GMTI data and evaluated as a tool for the web-