Robust image sparse matching method for high-precision aerial survey, storage medium and unmanned aerial vehicle

The invention belongs to the technical field of unmanned aerial vehicles, and discloses a high-precision aerial survey robust image sparse matching method, a storage medium and an unmanned aerial vehicle. The method comprises the following steps: weakening the mutual interference between the number of pixels and the gradient amplitude in an original SIFT matching algorithm by adopting a high-low frequency image information separation method, and respectively enhancing the contribution to the main direction; restricting effects in perspective transformation; enhancing the descriptor by using the image structure information, and processing the locally shielded image by using an anisotropic geodesic distance weighting method; utilizing a BBF (Best Bin First) algorithm and a parameter model method for carrying out feature point matching and mismatching elimination, and obtaining a final sparse matching point cloud through sparse bundle adjustment. According to the method, the obtained correct matching number is o