As a fine star-field identification algorithm, triangle algorithm is used far and wide currently, but there are some defects in triangle algorithm, such as low search efficiency and high mismatches probability. In allusion to these defects, a new triangle algorithm based on uncertain sign is presented. This algorithm extracted F and R features of star triangle, and then built a guidance characteristic catalogue which was searched by means of k-vector, promoting the search efficiency, moreover, in order to avoid the occurrence of mismatch, this algorithm would verify guide star triangle’s auxiliary information if its uncertain sign is 1. Simulation shows that: compared to the traditional triangle algorithm, this algorithm has a couple of advantages, including the higher rate of correct star recognition, lower mismatches probability, and better real-time adaptability and robustness. And this algorithm can reach 97% on identification rate when the position error is 2 pixels, and average identification time is 38.74ms; the traditional algorithm is 75% when the position error is 2 pixels, and average identification time is 187.26ms.