A target tracking algorithm based on mean shift and normalized moment of inertia feature

The paper presents a target tracking algorithm which is based on the mean shift algorithm and the normalized moment of inertia (NMI) feature, because the result of the moving target tracking in the air is not satisfactory by the traditional mean shift tracking algorithm. The NMI feature of the target is introduced, and the studied tracking strategy based on the minimum principle of the false alarm probability and the two-stage decision threshold of the similarity is constructed in the algorithm. The Kalman filter is used for the estimation and prediction when the target is occluded. The experimental results show that with the algorithm even the moving target in the air is in large deformation and occlusion, the system can effectively guarantee that the tracking is real-time and stable.