融合卫星遥感图像与航拍实时影像的拒止定位系统

A Denial-of-Service Resistant Positioning System Integrating Satellite Remote Sensing Imagery and Real-Time Aerial Images

摘要: 在卫星应用中，电磁干扰、卫星盲区、轨道受限等卫星拒止情况，可能会导致卫星依赖型定位方案失效，对生产、生活，甚至国防安全产生重大影响。面对卫星拒止定位的挑战，可结合无人机等航拍影像灵活机动的优势，为卫星遥感定位提供有效补充。融合卫星遥感图像与航拍实时影像构建拒止定位系统，核心关键在于实现两类影像的精准关联匹配，然而无人机航拍受飞行姿态波动、复杂地形影响，影像易存在大角度旋转、尺度偏移等几何变换，直接影响定位精度与可靠性。针对这一问题，本文设计了一种融合卫星遥感图像和航拍实时影像的拒止定位系统，并在该系统中嵌入了本文新提出的高精度图像匹配智能算法，该算法基于主导方向的图像匹配，保障了卫星拒止场景下两类影像的精准关联。实验结果表明，所构建的拒止定位系统在城镇、森林、沙漠、海洋四类典型地形场景中，针对航拍影像与卫星遥感影像的定位精度表现优异。该系统有效破解了卫星拒止条件下的定位难题，同时结合最先进的深度学习图像匹配算法，充分发挥卫星遥感图像的基准支撑作用与航拍影像的实时补充优势，为应急测绘、地理国情动态监测等卫星依赖度高的场景提供了高可靠、高效率的定位解决方案。

Abstract: In satellite applications, satellite denial scenarios such as electromagnetic interference, satellite blind zones, and orbital constraints can lead to the failure of satellite-dependent positioning schemes, exerting a significant impact on production, daily life, and even national defense security. To address the challenges of positioning under satellite denial conditions, the flexible and maneuverable advantages of aerial imagery from unmanned aerial vehicles (UAVs) and other platforms can be leveraged to provide an effective supplement to satellite remote sensing positioning. The core of constructing a denial-resistant positioning system by integrating satellite remote sensing imagery and real-time aerial imagery lies in achieving precise correlation and matching between the two types of imagery. However, UAV aerial photography is susceptible to flight attitude fluctuations and complex terrain, which often result in significant geometric transformations such as large-angle rotation and scale offset in the imagery, directly affecting positioning accuracy and reliability. To tackle this issue, this paper designs a denial-resistant positioning system that integrates satellite remote sensing imagery and real-time aerial imagery, and embeds a newly proposed high-precision intelligent image matching algorithm within the system. This algorithm, based on dominant direction-based image matching, ensures the precise correlation of the two types of imagery in satellite denial scenarios. Experimental results demonstrate that the constructed denial-resistant positioning system exhibits excellent positioning accuracy for aerial imagery and satellite remote sensing imagery across four typical terrain scenarios: urban, forest, desert, and marine. The system effectively solves the positioning problem under satellite denial conditions. Furthermore, by incorporating state-of-the-art deep learning-based image matching algorithms, it fully exploits the benchmark support role of satellite remote sensing imagery and the real-time supplementary advantages of aerial imagery, providing a highly reliable and efficient positioning solution for scenarios with high satellite dependency, such as emergency mapping and dynamic monitoring of geographical conditions.