Echtzeit-Qualitätsüberprüfung für zuverlässige UAV-gestützte Bilddatenerfassung und exakte, automatisierte Mehrbildauswertung

Photogrammetric systems are used for several tasks since decades. The development of small and lightweight Unmanned Aerial Vehicles (UAVs) that are suited to carry a consumer-grade camera, in conjunction with fully automatic 3D reconstruction systems, supports applications such as recurrent topographical surveys in open pit mining. In this article we describe that a fully automatic photogrammetric system can be used to measure distances
on a quarry wall using images acquired by an octo-rotor helicopter. To make the reconstruction more reliable in terms of accuracy and completeness, we propose a method that allows to assess already during the flight if the acquired images are sufficient to achieve the required properties of the offline photogrammetric reconstruction. This method provides online feedback to the user such that he or she can adopt the image acquisition strategy to obtain
a complete and accurate reconstruction. We show that based on these images a fully automatic reconstruction pipeline is able to obtain object points with an uncertainty of less than 0.1% of a 100 m quarry wall without the aid of external ground control points.