精细导星星库是支撑未来大口径空间望远镜观测的必要组成部分，是用来确定空间望远镜视轴绝对指向的关键依据。导星星库的容量和星库的均匀性对于实现精细导星功能和性能指标都极为重要。为了有效提高星图识别性能，提出基于变异系数的优选筛星算法进行导星筛选并构建星库。经过仿真和均匀性评价准则验证，基于变异系数优选算法构建的精细导星星库分布均匀性好，能降低冗余度，有效提高识别概率。

Fine guidance star catalog, a necessary part of the guiding process carried by large space telescope, is the key basis for the absolute pointing of the LOS of spacecraft. The guide star catalog capacity and star database uniformity are very important to realize the function and performance of the fine guidance sensor. In order to improve the performance of star pattern recognition, a new algorithm based on the coefficient of variation is proposed for guide star selection. The algorithm is proved better to solve the problem of navigation star uniform distribution by simulation and uniformity evaluation criteria, and the matching rate can also be greatly improved as long as the star pattern recognition rate