A Hybrid Adaptive Filtering Method for Target Tracking With a Radar and Imaging Sensor

摘要

Multisensor systems can produce more accurate estimation of target states. But tracking a maneuvering target is still a problem for multisensor tracking systems. This paper reports a hybrid adaptive filtering method, which combines variable dimension filter (VDF) with interacting multiple model (IMM), for a multisensor system with a radar and imaging sensor. An image-based maneuver detector using a feature vector extracted from the image performs better than the point-mass-based residual maneuver detector. When the maneuver detector declares an onset of maneuver, this technique uses an IMM estimator instead of a single model filter as used in the VDF algorithm. Simulation results show that the whole performance of the proposed method is better than IMM and VDF.