A detection adaptive filtering algorithm with interacting multiple model for multisensor tracking

A maneuvering target tracking algorithm for a multisensor system using an image sensor and other different sensors is presented; it combines a variable dimension filter (VDF) with an interacting multiple model (IMM). This technique uses an IMM estimator instead of the single model filter used in VDF when the maneuver detector declares an onset of maneuver. The performances of this approach, and of the VDF and IMM algorithms are investigated and compared. During a nonmaneuvering period, this approach is better than the IMM algorithm and equal to the VDF algorithm. During a maneuvering period, especially right after the end of a maneuver, this approach provides better estimations than the VDF algorithm.