Space observation images registration for dim target detection

Abstract

In this paper, we focus on the need of space observation images registration, in order to detect dim space targets more effectively. The global motion parameters are calculated first, using centroids of high light stars, which are corresponding during the entire image sequence, as feature points. Then, an algorithm of star image sequence registration based on Star Centroid Coordinates Matrix (SCCM) is proposed. The registration algorithm includes 3 steps: Firstly, star image is converted into SCCM. Secondly, utilizing global motion parameters, SCCM is processed. And thirdly, the star image sequence is registered, and at the same time, stars in the image are filtered. This algorithm can overcome disturbance from stars effectively. During the registration process, converting star image into SCCM can leave out image re-sampling and transformation steps, and can decrease computation complexity. Because stars are all filtered in the registered image, dim space target detection can be easier.