基于星联合分布的星图识别  

Star image recognition based on mutual star distribution

作　　者：姜笛[1,2,3] 张科 DEBEIR Olivier[2] 

机构地区：[1]西北工业大学航天学院,陕西西安710072 [2]法语布鲁塞尔自由大学图像、信号处理及声学实验室,比利时布鲁塞尔1050 [3]西北工业大学航天飞行动力学国家级重点实验室,陕西西安710072

出　　处：《系统工程与电子技术》2018年第1期127-131,共5页Systems Engineering and Electronics

基　　金：海装预研项目资助课题

摘　　要：针对栅格类算法中难以合适选取栅格尺寸的缺点,设计了将星图划分为多个多边形的划分方法,使星图中所有星都位于与其唯一对应的多边形中,提取多边形几何特征,对导航星库和观测星图的特征模式进行匹配。根据仿真数据进行验证,表明这一方法避免了栅格类算法中可能使多颗星位于同一栅格中的现象,同时也克服了栅格类算法不能正确选取定位星进而使星图识别失败的缺点,有效提升了星图识别的准确度。A novel method is designed to divide the star image into polygons,in order to solve the problem of setting the grid size in the grid algorithm.Via the method,each star is located in a corresponding polygon.The geometric features of these polygons are extracted and applied for matching the feature patterns of the guidance star database and the observed star image.Simulations indicate that the novel method removes the possible phenomenon in the grid algorithm where more than one star are located in the same grid.The grid algorithm's disadvantage of recognition failure resulting from identifying incorrect close neighbor star for reference star can be overcome and the accuracy of star image recognition is greatly improved.

关 键 词：星图识别 垂直平分线 多边形 特征模式 

分 类 号：V448[航空宇航科学与技术—飞行器设计]