Collaborative and Persistent Target Tracking and Acquisition (54348-CI)

Abstract

Abstract : Extending sensing capacity in target acquisition and tracking is very important for Army's future force. Video sensors provide both opportunities and challenges. One on hand, they bring rich and detailed information to facilitate identifying and recognizing the targets. But on the other hand, such visual information is subject to large uncertainties and ambiguities induced by factors such as cluttered and distractive backgrounds, illumination changes, visual occlusion, and low image quality. All these difficulties have impeded the development of effective and robust methods for persistent target tracking and acquisition in unconstrained environments. As a result, it is still challenging while desirable to have long-duration target tracking systems in many emerging applications, e.g., video surveillance and sensor networks. The goal of the proposed project is to overcome the challenges that confront persistent target tracking and acquisition, by developing new and effective computational models and methods that bridge computational sensing and some aspects in human perception of visual dynamics, and by creating multi-level synergy among various modalities and sensors.