三角形识别算法是采用星敏感器为导航部件所广泛运用的星图识别算法。但由于三角形特征维数较低，容易造成冗余匹配和错误识别。为了提高三角形识别算法的识别成功率，对三角形识别算法的识别过程做了改进。先将满足三边判决门限的导航星对记录到匹配数组，然后对导航星出现的次数进行计数；利用三角形三边之间两两相交的相关性，将导航星出现次数＜2的星对作为整体剔除，最后在余下的导航星对中寻找与观测三角形同构的三角形。识别过程中，以星对角距和相对星等差为识别特征，并对识别特征构造了散列函数，从而减少了识别过程中特征量比较的次数，提高了识别速度，增加了识别成功率。同时，也减小了内部导航星库的存储容量。

Triangle identification algorithm for star images has been widely used in attitude detection systems that take star sensor as navigation parts at present.However,its low characteristic dimensions lead to a higher redundancy and a lot of mistake identification. To improve the accuracy of triangle identification algorithm, a modified algorithm based on traditional one was proposed. By the proposed algorithm,the guide star pairs that meet interstellar angle threshold were recorded in a star matrix and the times appeared of the guide starts were counted simultaneously. According to the truth that each star must be attached to two sides of the triangle, the special star pairs that appear less than twice should be eliminated, then the guide stars for contructing triangles were searched from stars left in the star matrix. The algorithm takes the interstellar angle and relative magnitude of stars as the identification feature and also take the hashing search method into the star identification, therefore , it reduces the comparison times and increases the identification speed. Furthermore, it also decreases the capacity of guide star database.