An Autonomous Star Recognition Algorithm with Optimized Database

A novel star pattern recognition algorithm is presented for satellite attitude determination in the "lost in space" mode. The proposed method improves the speed in searching a large star catalogue by arranging it using a search tree data structure. The algorithm also processes the star image at grid level instead of pixel level to further improve the processing time. The experimental results show that the proposed approach significantly reduces the average run-time by 50% as compared with the conventional methods while still achieving slightly better star recognition accuracy at 95.07%.