中文摘要：

针对火星无人机探测飞行过程的特点及其机载计算机的局限性,在充分研究了矩阵奇异值向量性质特点的基础上,对奇异值向量进行主分量分析,提出了一种应用于火星无人机平飞段的基于奇异值分解的分层快速景象匹配算法,并给出了与之相应的机载特征数据存储方法。与相关算法的对比性实验表明,本文提出的算法具有准确,稳定,且速度更快,数据量更小的优点。通过仅在飞行末段,将本文算法切换成现有的基于S IFT算子的匹配算法,能在实现火星无人机全程快速景象匹配的同时,有效降低对其机载计算机综合性能的要求。

英文摘要：

Mars unmanned aerial vehicles（UAV） are one of the development directions of future deep space exploration.Through describing the characteristics of Mars UAV exploration flight and the limitations of its on-board computers,and on the base of full study of the nature of singular value feature vector,the principal components of singular value feature vector are analyzed.Then,a stratified fast image matching method based on singular value decomposition and its on-board data storage mode are proposed for level flight segment of Mars UAV.Simulation comparied with other correlation methods show that the method is more feasible.Using the SIFT operator-based matching algorithm at the end of the flight,switching between the two methods can reduce general performance requirements of its on-board computer,while achieving fast image matching for the whole Mars UAV exploration flight.