Absolute Visual Localization for Unmanned Aerial Vehicles

This book provides a systematic survey of unmanned aerial vehicle (UAV) visual positioning, with a primary emphasis on two central technical paths: image retrieval and image matching. These approaches lie at the heart of UAV–satellite cross-view geolocation, enabling accurate absolute positioning when GPS is unavailable or unreliable. We highlight advances in feature-based matching pipelines and deep learning–driven retrieval frameworks, which have become the most active and promising research directions. Other approaches—such as multi-sensor fusion, semantic and geometric modeling, and end-to-end learning frameworks—are briefly introduced to provide context, illustrating how UAV visual positioning is evolving into a broader ecosystem. By framing image retrieval and image matching as the core methods while situating them within the wider methodological landscape, this survey clarifies current progress, identifies open challenges, and points to future research opportunities. The intended audience includes graduate students, researchers, and engineers working on UAV navigation, computer vision, robotics, and geospatial information systems.