Marco BLOCK-BERLITZ1 / Benjamin DUCKE2
(1University of Applied Sciences Dresden, / 2German Archaeological Institute (DAI), Berlin, Germany)
Keywords: UAV, SfM, 3D recording, low-cost
Abstract:
The use of UAVs and image-based processing in archaeology shows signs of a technology in transition. At this stage, we can differentiate between a traditional “photogrammetric” approach that favours expensive camera systems, calibrated lenses and high resolution, high quality imagery, and a radically new “computer vision” approach that attempts to make the most of consumer grade hardware, massive input data and highly efficient, often GPU-based software. Both approaches deserve to be explored for their individual strengths and weaknesses.
This Paper focuses on the “quantity beats quality” approach, presenting a range of case studies where cheap and sturdy UAVs equipped with equally cheap and robust camera systems, using distorted but efficient fish-eye lenses, have been put to the test in diverse, sometimes extreme environments. We present advances in using highly automated processing methods to generate detailed 3D models and derivative products such as terrain models and orthophotos from regular HD video streams instead of high resolution single shots.
Constant progress in computer vision algorithms means that we can get ever more out of our data, and our case studies will demonstrate why we think that money is better spent on lots of cheap consumer hardware then on a few expensive systems. The cheaper the devices, the more obvious the discrepancy between the necessary investments for hardware and software licenses. Therefore, we will showcase the kind of results that archaeologists can expect from software that is available at no cost, be it as freeware or open source software.