New tracking algorithm based on the nearest neighbor star search approach is proposed for fast star identification. In order to improve the performance of tracking, this algorithm is composed of two parts. One is a searching part of unknown stars in the field of view (FOV) using nearest neighbor star search approach. The feature of this method is that each star has connections with some adjacent stars. Unknown stars can be identified by tracing the nearest star from the previously recognized star with reference to the star catalog. Star neighborhood information is a list of neighbor stars in the order of closeness and included in the catalog. The other is a predicting part of position of stars on the current image frame. Star position can be predicted from satellite angular velocity and previous attitude information. In this technique, angular velocity is estimated by the last two captured images without gyroscope observation. Most of the stars in the FOV are tracked properly by matching star centroid position on the captured image with the predicted star position. Star trackers for micro-satellite developed by Tohoku University so far had only lost-in-space attitude determination algorithm, whose operation frequency was limited down to 1 Hz. It is expected that the above mentioned star identification method enables improvement both in reliability and operational frequency. This algorithm was evaluated in PC simulation. The results show that attitude determination can be carried out over twenty times faster compared to the conventional method. Hence, it is illustrated that the efficiency of star identification is improved by these approaches.
Development of fast tracking algorithm using nearest neighbor star search approach
2016-03-01
392944 byte
Conference paper
Electronic Resource
English
Fast nearest-neighbor search algorithm [2669-11]
British Library Conference Proceedings | 1996
|Fast multidimensional nearest neighbor search algorithm using priority queue
Online Contents | 2008
|Fast Algorithm for Nearest Neighbor Search Based on a Lower Bound Tree
British Library Conference Proceedings | 2001
|Nearest neighbor search for relevance feedback
IEEE | 2003
|