Cross-view image matching method for unmanned aerial vehicle image and satellite image

Abstract

The invention discloses a cross-view image matching method used between an unmanned aerial vehicle image and a satellite image, and belongs to the technical field of remote sensing images. According to the method, after the similarity between the unmanned aerial vehicle visual angle image and the satellite visual angle image is considered, the images of the two visual angles of the same target place are regarded as the same class of images without difference, namely the place is regarded as the class, and place classification models under the two visual angles are trained. The classification models without classification layers are taken as image feature extractors; based on the image feature extractors, feature extraction is carried out on the images requested to be matched and each imagein a candidate image library to obtain an image feature vector of each image; based on the image feature vector, the image feature similarity between the image requested to be matched and each imagein the candidate image library is calculated, and the highest image feature similarity is taken as a matching result of the image requested to be matched. The method can be used for image-based unmanned aerial vehicle image positioning and unmanned aerial vehicle navigation.

Description

technical field [0001] The invention belongs to the technical field of remote sensing images, and in particular relates to a cross-view image matching method used between drone images and satellite images. Background technique [0002] Remote sensing technology has developed rapidly in recent decades, and remote sensing platforms have gradually become diversified. Satellite platforms, aviation platforms, and UAV platforms have all become important remote sensing image acquisition platforms. In particular, UAVs have been widely used in various industries because of their advantages such as strong mobility, little influence from weather, and low cost. In most cases, UAV image positioning and UAV navigation can be well solved by GPS. But one still faces the problem of locating drone images without geotagging and navigating without GPS. Cross-view image matching has received a lot of attention in recent years because it can achieve image-based geolocation. Its main content i...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to: aim at the technical problem that the existing ground and satellite image (aerial image) cross-view matching method is not suitable for the cross-view image matching process between the UAV and the satellite view, based on the UAV image The similarity with satellite images, using the idea of ​​location classification, realizes the mutual matching of satellite perspective images and UAV perspective images

Images