Cluster unmanned aerial vehicle real-time dense reconstruction method

According to the cluster unmanned aerial vehicle real-time dense reconstruction method, relative positioning is carried out based on GNSS and VIO, a plurality of unmanned aerial vehicles carrying cameras, IMUs and GNSS are used for scanning a target area, and cooperative reconstruction is completed. The relative positioning mode is realized through a dynamic distributed network DDN. The DDN uses prior relative position information to adjust the structure of the distributed network in real time, and for each sub-module of the distributed structure, accurate local relative pose information is used to complete dense reconstruction. In each sub-module, a center node is set, and the pose between virtual center nodes of each sub-module is estimated to fuse the point cloud. And point cloud registration is carried out between the unmanned aerial vehicles in the sub-modules and between different sub-modules of the whole distributed network by adopting a layered point cloud fusion algorithm, so that a registered high-p