As the traditional triangle star identification algorithm is insufficient, this paper proposed a fast all-sky autonomous triangle algorithm with star magnitude-independent. By structuring the maximum interior angle and two sides as a matching feature triangle, the algorithm established celestial navigation feature library which was constructed to be a hash function according to the maximum interior angle and stored into sub-blocks. ‘Edge-angle-edge’ matching mode was adopted in the process of star identification. First, adopt the hash search to achieve rapid positioning of sub-blocks in terms of observations of the maximum interior angle, and then quick matching of star argument was conducted on both sides of the observed interior angle, which would further to narrow matching scope of the argument and improve identification speed. Experiments indicate that identification rate of the algorithm can exceed 98.08% when star point noise is lower than 2 pixels and average identification time is 13.1 ms when observed stars number equals to 10 and the sum of sub-blocks in feature library is 1 024. Compared with current triangle identification algorithms, this algorithm has obvious advantages in identification speed, identification rate and the ability of resisting star magnitude noise.