Target tracking using Kalman Filter Embedded Trust Region

This paper proposes a novel algorithm, the Kalman Filter Embedded Trust Region (KFETR), for target tracking. Kalman filter and trust region are two successful methods for object tracking. The presented KFETR algorithm integrates the advantages of the two approaches. The new algorithm makes full use of the target's moving information and predicts the target's approximate position firstly. Because of these properties, the algorithm overcomes the problem that trust region converges to a local minimum which is not of interest caused by improper initial position. Promising experimental results on several image sequences demonstrate the robustness and effectiveness of KFETR.