2014 · 13 citations · 4 references
ComputingComputer ScienceComputer VisionMachine VisionSensor FusionBayesian FilteringKalman FilteringManeuveringState-space ModelsTracking SystemsVisual TrackingAdaptive Kalman FilterInteracting Multiple Model FilteringKalman Filter TrackingSurveillance SystemsTarget TrackingTransition Probability MatrixUnscented Kalman FilterUnscented Kalman Filtering
Maneuvering target tracking is very important in visual surveillance systems. However, as a popular algorithm, the standard interactive multiple model (IMM) filter has imprecise estimation and computerized intractability. To solve the problems, an improved IMM filter is proposed in this work. First, the Kalman filter (KF) and the unscented Kalman filter (UKF) are employed to estimate the non-maneuvering and maneuvering motion states respectively. Above all, the transition probability is adaptively calculated by using the posterior probability. Numerical simulations are presented to compare the tracking performance of the proposed filter with that of the standard filter. Finally, the results show that the proposed filter is competent for the maneuvering target tracking with excellent performance.
4