Spratling, Benjamin (2011-05). Star-ND (Multi-Dimensional Star-Identification). Doctoral Dissertation. Thesis uri icon

abstract

  • In order to perform star-identification with lower processing requirements, multi-dimensional techniques are implemented in this research as a database search as well as to create star pattern parameters. New star pattern parameters are presented which produce a well-distributed database, required by the database search algorithm to achieve the fastest performance. To mitigate problems introduced by the star pattern selection, incorrect entries are added to the database, which reduces the number of iterations of the run-time algorithm. The associated algorithms, star pattern parameters, and database preparation are collectively referred to as Multi-dimensional Star-Identification (Star-ND). The star pattern parameters developed may also be extended to star patterns with an arbitrarily large number of stars, while retaining the well-distributed property. The algorithm is contrasted with the current state-of-the-art star-ID algorithm, Pyramid. The database is found to grow linearly with the size of the star catalog, while Pyramid's database grows quadratically. The running time of Star-ND is found to be on average a factor of 25 times faster than the time for Pyramid.
  • In order to perform star-identification with lower processing requirements, multi-dimensional techniques are implemented in this research as a database search as well as to create star pattern parameters. New star pattern parameters are presented which produce a well-distributed database, required by the database search algorithm to achieve the fastest performance. To mitigate problems introduced by the star pattern selection, incorrect entries are added to the database, which reduces the number of iterations of the run-time algorithm. The associated algorithms, star pattern parameters, and database preparation are collectively referred to as Multi-dimensional Star-Identification (Star-ND).

    The star pattern parameters developed may also be extended to star patterns with an arbitrarily large number of stars, while retaining the well-distributed property. The algorithm is contrasted with the current state-of-the-art star-ID algorithm, Pyramid. The database is found to grow linearly with the size of the star catalog, while Pyramid's database grows quadratically. The running time of Star-ND is found to be on average a factor of 25 times faster than the time for Pyramid.

publication date

  • May 2011