Image-Based Tracking for Smart Muintions

In this proposal we consider the development of statistical image processing, tracking, and guidance technologies for navigation, guidance, and control in advanced smart munitions. The archetypal example system in which we are interested is the LOCAAS smart munition. The algorithms we propose to develop, however, will be broadly applicable to any number of seeker, interceptor, and semi-autonomous munition systems. Leveraging recent research in optical flow and active contours with proven filtering technologies of direct applicability in tracking for systems such as the Air Borne Laser, we will develop a suite of algorithms to determine feature positions over time within scenes observed by imaging-sensor-bearing munitions. The feature positions over time are then to be used for improved track and guidance accuracy. Much more general and flexible than traditional correlation trackers, the optical flow methodology adapts well to scaling and aspect angle changes. Coupled with efficient temporal models and filters, the optical flow and active contours can maintain the location of multiple targets in a scene with great accuracy.