New star identification algorithm based on Hausdorff distance for the initial attitude establishment mode in star sensor

One of the best attitude sensors for space applications is the star sensor. This sensor uses the stars in the field of view for attitude determination. The star sensor can determine the attitude with and without knowledge of the previous attitude. Determining the attitude without knowledge of the previous attitude in the star sensor is called the initial attitude establishment mode. The star identification algorithms of this mode are called the lost-in-space algorithms. Because in this mode, all the stars of the celestial sphere must be searched. This paper presents a new lost-in-space algorithm based on Hausdorff distance. Two different approaches have been proposed for identification. The first approach is designed based on the pivot star, and the second approach uses the segmentation of the celestial sphere. To increase the identification rate, another step based on a specific angle also has been examined. This step does not require a database. To select neighboring stars in these methods, three approaches have been proposed, one approach is based on star magnitude, and the two other approaches are based on two cost functions. The results show that the approach of using the pivot star along with the angle and selecting the neighboring stars based on the magnitude has a better performance and its identification rate is 99% for stars brighter than 6 \({\text{M}}_{\text{v}}\). Also, comparing the identification time of the Hausdorff algorithm with the pyramid algorithm, in the case of a linear search of the database, shows that the average identification time for stars brighter than 6 \({\text{M}}_{\text{v}}\) is 13.75 times less than the pyramid algorithm.