基于空间关系几何约束的无人机景象匹配导航  

Scene matching navigation for UAV based on spatial relationship geometric constraints

作　　者：李耀军[1] 潘泉[1] 赵春晖[1] 张绍武[1] 凌志刚[1] 

机构地区：[1]西北工业大学自动化学院,西安710072

出　　处：《计算机应用研究》2010年第10期3822-3825,3846,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60634030;60702066);国家教育部新世纪优秀人才资助项目(NCET-06-0878);航空基金资助项目(20090853013);西北工业大学校"翱翔之星"计划资助项目

摘　　要：针对GPS的信号强度较弱、易受各种电磁干扰,提出一种基于空间关系几何约束的景象匹配导航算法。首先基于空间关系几何约束的多匹配区选择方法,将实时图划分为多个分区;然后采用基于边缘响应的加权Hausdorff距离景象匹配算法对各个匹配区进行定位计算;最后,通过最优配准点坐标估计策略,解算出无人机实时图中心在基准图上的精确定位坐标。多区域景象匹配采用并行计算方法,利用历史导航信息来辅助修正景象匹配导航误差。实验结果表明,该算法可较好地满足无人机对景象匹配导航算法实时性、精确性的性能要求。The GPS system’s signal was extremely weak and was vulnerable to both intentional and unintentional interference. For the complexity of the work environment of scene matching navigation on the platform of UAV,this paper proposed a multiregion scene matching algorithm for suitable-matching area based on the spatial relation constraint. This algorithm adopted parallel computing to improve the real-time performance of SMNS. It could estimate ultimately the optimal allocation of coordinates of center point by calculating the spatial relation of many match areas. So it could get the precise location of real-time image center at the base map. At last,designed a method based on previous navigation data to check the location. To accomplish the on-line scene matching reliably and precisely,this algorithm only needs one real-time frame image to match base map every time. Experiments show that this algorithm meets the UAV navigation performance requirements such as real-time,robustness and high accuracy.

关 键 词：空间关系几何约束 加权HAUSDORFF距离 并行景象匹配 历史导航信息 无人机 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]