Automatic star pattern recognition in tracking mode

Autonomous star trackers are rapidly becoming the most accurate means of attitude determination due to their high accuracy, small size, light weight and simple functionality, making them devices of choice in nearly all modern space vehicles. In all star trackers star centroiding and pattern recognition are fundamental processes, necessary for attitude determination. In this paper an automatic star pattern recognition scheme is presented for currently available high frame rate imagers especially for Active Pixel Sensors (APS) (because of their random access ability) in tracking mode. This scheme uses the knowledge of boresight direction to pick stars from star catalog and convert them in imager's coordinates to predict the regions of star locations in next image frame. By keeping track of the reference catalog stars and centroiding results of the predicted stars regions from image, the pattern recognition is done automatically. The boresight direction is updated at each sequential frame. Simulation results show that the proposed scheme is computationally more efficient and faster as compared to normal centroiding scheme. In addition, it is shown that it can easily handle the problem created by double stars and it inherently decreases the probability of false stars and noisy spikes.