Star Identification Algorithm Based on Log-Polar Transform

A star identification algorithm based on Log-Polar transform is proposed in this paper. First, Log-Polar transform is introduced to generate star patterns. Second, the generated star patterns are coded to strings for the purpose of quick match and memory saving. Finally, a modified string match algorithm based on Knuth-Morris-Pratt is used to find the matches between measured stars and guide stars. In simulations, using the stars brighter than magnitude 6.0 and a 12 x 12 degree field of view, the algorithm obtain an identification rate of 98% from the statistics of 1000 random sensor orientations at a positional noise level of two pixel. Meanwhile, the memory usage of the algorithm is comparatively small.