UAV Localization Based on Signage Text Recognition and Matching with POI

Localization is the most important basic information for UAVs during their missions, but within cities, the multipath effect formed by high-rise buildings will significantly reduce the positioning accuracy of GNSS. To achieve autonomous UAV localization in low-altitude scenes within cities, we propose a UAV image scene localization method based on signage text recognition and matching. First, extracting the text in store signages from UAV images, then fuzzy matching them with the POI names in electronic maps, finally, achieving scene localization of UAV images using multiple POIs jointly. The distance between image scene position and UAV flight trajectory in flight experiments is no more than 40 meters, which is consistent with the actual situation. Therefore, this method has good autonomous localization capability under urban low-altitude conditions.