Fast Star Pattern Recognition Using Spherical Triangles

Abstract

A current method by which star trackers identify stars is to match the angles between stars within its field of view to angles stored in a catalog. If an angle can be matched to one pair of stars, the attitude of the star tracker can be determined. However, the measurement of the angle will include error, and so the true angle can only be known to lie within a certain measured range. The result is, after comparing the measured angle to the catalog of angles, more than one pair of stars can be the correct solution. A method for narrowing down to one solution involves employing many angles within the field of view in a certain order, called “pivoting,” which can be time consuming and does not always yield a solution. The method presented here matches spherical triangles made from sets of three stars within the field of view to spherical triangles stored within a catalog. By using both the area and polar moment properties of the spherical triangle, the range of possible solutions is very quickly narrowed, fewer pivots to other spherical triangles are required, and the method is more likely to yield the correct solution than the angle method.