Gaussage and online parameter based video tracking mode transition from KF to PF for optimal performance

To tackle the problem in which one currently executing algorithm cannot provide its optimal performance any more during a given video target tracking scenario, another designated algorithm takes over which carries out its best performance providing the conditions are satisfied. In this paper, we continue the previous research by investigating the conditions in which the video target tracking transits from the particle filter (PF) mode back to the Kalman filter (KF) mode. First, the switchover of the KF-to-PF mode is briefly reviewed, the definition of Gaussage and online performance evaluation parameter are introduced and applied first in the simulation for the transition of the PF-to-KF mode, then the Gaussage and the parameter are employed in a given real video target tracking process to test the PF-to-KF mode transition. The result suggests the feasibility of the proposed tracking mode transition method.