Star spot extraction method for motion blurred star image assisted by prior information

Abstract:

In highly dynamic environments, there can be motion blur on the star images taken by star sensors, which will seriously affect the star spot extraction accuracy. Regarding to such issues, a method is proposed in this paper to directly extract star spots form motion blurred star images assisted by prior information. First, the prior attitude information of inertial navigation systems and star sensors was utilized to predict the coarse star spot coordinates on the motion blurred star image, which were regarded as seed points for local region growth to obtain the range of the star spots. Then, the star spot motion trajectory was estimated based on the angular velocity output of the inertial navigation system, and star spot coordinates were calculated by the centroid method. Finally, errors of the star spot coordinates were corrected according to the relationship between the star spot motion trajectory and the centroid extraction error. Simulation results show that compared with the method that extracts star spots after restoring the motion blurred star image, the proposed method can ensure high accuracy of star spot extraction and good anti-noise performance under highly dynamic environments. Besides, the proposed method can greatly reduce the processing time for motion blurred star images, which is helpful to ensure that the data update rate of star sensors is not greatly affected by highly dynamic environments.