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

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.