A NEW APPROACH TO POINT-PATTERN MATCHING

We describe a new algorithm for matching star lists, given by their two-dimensional coordinates. Such matching should be unaffected by translation, rotation, rescaling, random perturbations, and some random additions and deletions of coordinate couples in one list relative to another. The first phase of the algorithm is based on a characterization of a set of coordinate couples, relative to each individual coordinate couple. In the second phase of the algorithm, the matching of stars in different lists is based on proximity of feature vectors associated with coordinate couples in the two lists. The order of magnitude computational complexity of the overall algorithm is n2 for O(n) coordinate couples in the coordinate lists.