Fast Access for Star Catalog Based on Locality-Sensitive Hashing

Abstract

In order to improve the access speed and robustness of star catalog database during star identification, an algorithm based on locality-sensitive hashing is proposed. First, according to principle of star identification, the angle distances are quantified on the basis of angle distance error limit, and the order star set pattern is transformed into an array of integers, which has locally sensitive hash characteristics. Then key of hash obtain by hashing the array of integers with Stlport, and the value of hashing is a set of central star number in the ordered star point set pattern. Numerical simulation results indicate that the time complexity of proposed algorithm is O(1), which is much better than direct search, binary search and k-vector search technology. In addition, the proposed algorithm is robustness due to the affect is not significant as the performance influenced by angle distance error limit. Considering practical application, the error limit of angle distance could be chose as 1 pixel, the number of quantified angle distances could be chosen as 3. Under this condition, the collision rate in hashing table is 0.74%, the average searching time is 1.007 4 and the average consuming time is 22 μs during star identification.