一种星模式识别算法的稳健性研究  

Study on the Robustness of A Star Pattern Recognition Algorithm

作　　者：朱长征[1] 沈振康[1] 李飚[1] 

机构地区：[1]国防科技大学ATR国家重点实验室,长沙410073

出　　处：《宇航学报》2004年第4期434-438,共5页Journal of Astronautics

摘　　要：研究了一种基于奇异值分解的星模式识别算法的稳健性。首先简述了研究星模式识别算法的必要性和现有的星模式识别算法。从数学原理上证明了本文算法进行模式识别所用的矩阵的奇异值相对于坐标变换是不变的。从物理意义上给出了稳健的解释。仿真试验选取了四种有代表性的情况进行了研究。结果证实该算法所用的特征不变量是稳健的。结论部分分析了该算法的优缺点。最后给出了待研究的问题。In the article, the robustness of a star pattern recognition algorithm based singular value decomposition was researched. Firstly, the article briefly stated the need to study star pattern recognition algorithm,and some kinds of star pattern recognition algorithms in existence were introduced. Then the singular values of the two matrixes used for star pattern recognition algorithm in the article were proved to be invariant with respect to coordinate transformation in mathematical principle. The physical interpretation of the robustness was presented. In the simulation part,four kinds of representative situation were studied. The characteristic invariables used in the star pattern recognition algorithm are proved to be robust by simulation results. In the conclusion the advantage and disadvantage of the algorithm are analyzed. In the end, some issues to be studied are listed.

关 键 词：奇异值分解 模式识别 稳健 

分 类 号：V448.2[航空宇航科学与技术—飞行器设计]