Design And Implementation Of Fast Star Identification Algorithm

Navigation system is an indispensable equipment of the spacecraft,which has an important role in the work of the spacecraft.Accurate attitude information is the base of the spacecraft navigation system.The star sensor is a kind of high precision attitude information measuring instrument in aerospace field.Research on star identification,the key technology of star sensor,is of great significance for the development of aerospace industry.Star identification contains three steps:image acquisition and preprocessing,database construction and features matching.This paper proposes the improvements of star map recognition from the three aspects to improve accuracy base on the existing algorithms.To begin with,for image acquisition and preprocessing,a self-adaptive background prediction algorithm by variable neighborhood for star extraction is proposed based on the existing background prediction algorithm.At the edge of the Star point,the prediction neighborhood is composed by the pixels those grey value are smaller than the median in the neighborhood.While the fixed weight background prediction is used in other parts.Second,based on the noise properties of the pixels to be predicted,adjust the weight of the pixel to be predicted to predict the background adaptively.The simulation and calculation shows that the modified background prediction algorithm can suppress the background noise better and confirm Star point clearer than the conventional algorithms.Next,for database construction,a new optimal selection algorithm is proposed based on orthogonal grid method.By analyzing the distribution of the celestial stars,the new algorithm introduce a new concept,separation-magnitude weight to select navigation star.The new algorithm is more uniform than the existing orthogonal grid method.In addition,in approach to the double star problem,taking the equivalent of the double star into the star database to make the database more complete.On the third,for the features matching,an improved star pattern identification algorithm is proposed based on the existing triangle algorithm and its related improved algorithms.In this algorithm,a quadrilateral is formed by introducing auxiliary star outside the star triangle,a statistical method is proposed to identify the stars.Verify the recognition result by using features of the star triangle.Simulation results show that compared with the existing algorithms,the capability of Anti-interference is improved.At last,the main work and the innovative aspects in this dissertation are summarized,and the further research is prospected.