【摘要】 讨论了一种新的无须任何先验知识的星图识别算法。通过对导航星座数据库进行调整和最大限度地利用了已识别基元所提供的匹配信息，使得算法的计算量和存储容量与传统的三角形匹配算法相当。ＭｏｎｔｅＣａｒｌｏ 模拟表明，在星等误差０．５ 星等（３σ） ，位置误差１０ 角秒（１σ） 时，算法的识别成功率接近１００％ 。更多还原

【Abstract】 A new star pattern recognition algorithm without any a priori attitude knowledge is described. It uses an ordered star triplet defined by three stars and their celestial coordinates as recognizable unit, and makes up multiple units into a group. A group having maximum credibility is selected as correct recognition. By using special arrangement for the database of guidance patterns and the information provided by the recognized unit, the computation and the storage capacity of the on board star catalog are comparable with those of the traditional triangle matching. Monte Carlo experiment demonstrates that, while the magnitude uncertainty is 0.5 (3σ) and the position uncertainty is 10 arcsec (1σ), the success rate is very close to 100%.更多还原