A spline based model for incorporating prior knowledge of target routes in radar tracking

A spline based model for incorporating prior knowledge on target routes is presented. Splines form a generic tool for describing target trajectories. Also, they are easy to generate from sampled target data. The model is formulated as a nonlinear state model, and can thus be employed directly by a tracking algorithm based on the extended Kalman filter. An interactive multiple model filter with variable structure is used for tracking both the motion parameters, and if a target is on a particular route. The spline model and the multiple model filter are evaluated on a set of real target paths extracted from an air surveillance scenario.