Target registration assignment in the presence of uncertainty

In target tracking scenarios when multiple sensor platforms are present, there is the possibility that sensor registration will occur. If the problem is not properly compensated for, the combined target tracks will have a greater error as translation and rotation errors will be incorporated into the solution. In fact, some of the errors may result in some target tracks not being properly associated so that more targets will be seen in the combined solution. While many techniques exist to correct the registration error, they do not assign the target tracks from one sensor platform to those tracks of another platform. In this paper, a recently developed assignment technique for target registration is expanded upon to incorporate track uncertainty. The technique is based on a phase-only filter image correlation method. The track features can be mapped into an invariant image space that can be compared. The original technique did not consider target uncertainty. A new representation is considered in this paper.