Super-telephoto Drone Tracking Using HFR-video-based Vibration Source Localization

In this paper, we propose a super-telephoto drone surveillance system that can search and track a flying drone at a 1 km distance or more within an area spanning hundreds of square meters. By constructing a mirror-drive high-speed pan-tilt control system with a super-telephoto camera lens, remote drone surveillance becomes possible owing to the acquisition of high spatial resolution images of target objects. We adopted a vibration-based drone localization algorithm that can simultaneously detect fast-rotating drone-propellers in an high-frame-rate video using a pixel-level short-time Fourier transform to achieve robust surveillance under various challenging conditions for super-telephoto video-shooting such as low luminance and heat haze. The effectiveness of the super-telephoto drone tracking system is demonstrated by conducting a drone searching and tracking experiment at a 1 km distance using an 800-mm focal length lens with a mirror-drive video-shooting system.