A New Approach to the Electro-optical Tracking of a Space Object

The electro-optical tracking of a space object is often interrupted by the dropped frames under the interference of stars. In this paper, the two stages of the tracking process, namely the state extrapolation and real-time sampling, are studied from a probabilistic point of view. Based on the theory of estimation, a new method is proposed, which improves the robustness of prediction by utiliz- ing the data-fusion property of Kalman filtering, and realizes the adjustment of the tracking window by the mean shift technique. The corresponding experiment shows that this method can effectively reduce the impact of interferences, and improve the robustness of the tracking process.