A Random Point Process Approach to Multiobject Tracking

This paper presents a new theoretical approach to multiobject tracking problems which might provide a framework for developing effective analytical tools for multiobject tracking problems. The approach is based on a random spatial point process model of measurements and a Laplace transform method for calculating the likelihood formulas that are essential to deriving tracking algorithms and to analyzing tracking performance. The model generalizes the usual one, but it is much more compact to express and simpler to understand (once easy, but unfamiliar mathematics is grasped).