摘要
提出了一种新的星图识别方法,这种方法的基本思想是首先从星敏感器中提取全部星图信息,把提取的星图组成一个待识别数据集合B ,把全星图参考星库作为一个标准数据集合A ,然后,计算2个数据集合A ,B之间的相对Hausdorff距离,根据最小相对距离得到待识别星的位置,其中A ,B分别由矢量坐标a ,b构成.与传统的星光角距识别方法不同,这种方法充分利用了星敏感器中的所有星的空间结构信息,所以,它对各种噪声干扰具有很强的鲁棒性.半物理仿真实验结果表明,在具有强的随机噪声、星图畸变和一定数目的流星干扰情况下,这种方法具有很好的识别效果.
A star pattern recognition approach has been developed. Its based idea was that extracting the star pattern information, fitting together the information to a pending data set B, putting all star catalog information into a criterion data set A, and computing the relative Hausdorff distance of two sets, here a and b is a vector element of set A and set B. The star was recognized according to the minimum Hausdorff distance. The approach is different from the conventional star pattern recognition approach, which uses the starlight angle-distance method to get the position, it takes the most of the information of star dimensional configuration, therefore, the approach has strong robusticity for the disturbance of noise, distortion, and a few meteor. In the case of random noise, imaging distortion, and a few meteor disturbing, the experimental results of half physical simulation indicate that the approach has good recognition effect.
出处
《北京航空航天大学学报》
EI
CAS
CSCD
北大核心
2005年第5期508-511,共4页
Journal of Beijing University of Aeronautics and Astronautics
基金
国家自然科学基金资助项目 (60 1740 3 1)
遥感信息处理国家重点实验室资助项目 (WKL (0 2 ) 0 10 2 )
