Feature point set image matching algorithm for satellite attitude determination

This paper constructs a template for description of global image feature by integrating matching methods of grayscale and geometry feature, selecting feature point set depending on partial texture energy distribution and utilizing geometrical constraints among points. The matching between real-time image and reference image is realized through stepwise refinement method. The first step is a coarse search by feature point set matching. Fine registrations include cluster analysis and close object matching. Compared with traditional image matching algorithm, feature point set algorithm can improve matching speed and closed object matching can meet the precision requirement. Its practicability has been proved by simulated experiments