DiSCVer: Discrimination using Sensor Collaboration & Verification

SSCI proposes the development of DiSCVer (Discrimination using Sensor Collaboration and Verification), a multi-sensor Automatic Target Recognition (ATR) and Tracking system, based on a layered sensing architecture that will maximize the information available from multiple ISR assets to increase the situational awareness of decision-makers. At the core of DisCVer are two information fusion modules: a density tracker that will accept multiple sensor inputs to track movers in a scene, and a Learning Classifier System (LCS) which will use features extracted from multi-sensor data to perform dismount discrimination. Detection and feature extraction capabilities associated with specific sensors will provide the inputs to the two central information fusion modules. In Phase I, SSCI showed the feasibility of the system through the development of algorithms for detection and feature extraction from multiple sensor modalities, and for tracking of movers and dismount discrimation. The Phase II will build on our Phase I efforts with the goal of enhancing the performance and efficiency of individual detection, tracking and fusion components to create a single, integrated system for efficient, online dismount discrimination and autonomous tracking. At the end of Phase II, SSCI will deliver a prototype dismount tracking and discrimination software library for use in a layered sensing environment. BENEFIT: The DiSCVer system proposed in this effort will reduce the burden on decision-makers in Intelligence, Surveillance and Reconnaissance (ISR) operations by fusing information from complex sources together and presenting them with information in an efficient manner. Also, decision-makers will be able to task available assets to provide information that further improves their situational awareness of a mission. The automated algorithms will be of similar use to the government in Homeland Security applications by reducing the burden on personnel involved in monitoring and surveillance. The detection, tracking and information fusion algorithms proposed in this effort will also be of use in civilian applications such as disaster (natural or man-made) relief operations, environmental monitoring, medical engineering, etc.