Instantaneous Geo-Location of Multiple Targets From Monocular Airborne Video

Abstract

We propose a robust and accurate method for multi-target geo-localization from airborne video. The difference between our approach and other approaches in the literature is fourfold: 1) It does not require gimbal control of the camera or any particular path planning control for the UAV; 2) It does not rely on GIS information, or otherwise any geo-referenced terrain database, elevation map, or accurate altimeter for estimating the UAV's and target's attitudes; 3) It can instantaneously geo-locate multiple targets even if they were not previously observed by the camera; and 4) It requires only one camera, and it employs a virtual-stereo technique using the image sequence for increasing accuracy in target estimation. The result is a method that can reach a few meters of accuracy for an UAV flying at a few hundred feet above the ground. Such performance is demonstrated by computer simulation, in-scale data using model city, and real airborne video with ground truth.