Non-Cooperative Target Detection/Identification (ID)

Engagement and sensing tasks require the ability to quickly and accurately identify friendly and enemy targets. This rapid and precise ID information is useful particularly in situations where immediate targeting of enemy aircraft is necessary. Since cooperative techniques such as IFF do not always distinguish to the extent needed, non-cooperative techniques for detecting and identifying friendly/neutral/enemy targets are necessary. In particular, ways of fusing information provided by multiple sensors can improve detection/identification capability. Toyon Research Corporation proposes to analyze multi-intelligent data sources and to research a dual layer solution for non-cooperative target identification. The primary layer uses a Multiple Hypothesis Tracker (MHT) in conjunction with a Bayesian network to model feature information and possible inferences garnered from this information in a way that promotes improved measurement-to-track association. Toyon’s Tracked Object Manager (TOM) will handle feature database management and use its databases to correlate on-the-fly information for input into the Bayesian network. Toyon will also design a test scenario in order to test these algorithms. BENEFIT: The algorithms developed on this effort will support a layered sensor architecture in which multi-sensor data is fused to detect and identify objects that move around within the sensor network. This information can then be used to provide a clear uncluttered description of targets in a single integrated picture. Such a system has wide applicability to a variety of Intelligence, Surveillance and Reconnaissance and Security missions. Not only is it applicable to Air Force scenarios, it may also be applied to situations such as border security, using video surveillance systems which may use features such as those found by a scale-invariant feature transform algorithm.