Automated georeferencing and orthorectification of archaeological aerial photographs

Aerial archaeology is one of the oldest archaeological prospection methods. Although it is highly efficient in detecting sites, its full potential (i.e. the detailed mapping of archaeological and palaeoenvironmental structures) can only be used, when the photographs are (ortho)rectified, interpreted and mapped. Georeferencing of the aerial images is commonly determined with ground control points (GCPs), whose measurement and identification is a pretty time-consuming operation. It is therefore the biggest bottleneck impeding large-scale archaeological mapping projects with thousands of images. The general aim of the proposed project is therefore to create automatic solutions for the automated georeferencing and orthophoto generation of archaeological aerial photographs (APs). The resulting workflow should allow the efficient processing of both newly generated and previously acquired aerial images based on available reference data. In order to fulfil these requirements, the project will (1) evaluate a calibrated digital still camera (DSC) with an attached positioning and orientation system (POS) for archaeological prospection and automated orthophoto production = "hardware approach"; (2) develop a workflow including software for automatic orientation of APs without accurate initial positioning and rotation information = "software approach". Both approaches will be implemented in the conventional aerial reconnaissance work from a small aeroplane as well as in an innovative method using an unmanned aerial vehicle (UAV) to acquire aerial imagery completely autonomously and being independent on conventional aircrafts. An archaeological case study will serve to test and evaluate all proposed solutions for automated orthophoto generation. The proposed 250 sq km large area is largely covered by vertical and oblique APs from various years and seasons, with scales ranging from 1:25.000 to 1:5.000. The evaluation should identify problems and improve our approach, both using conventional aeroplanes as well as UAVs. The case study should also result in written guidelines about the methods of proper aerial data acquisition and georeferencing. Three international partner organisations (Department of Prehistoric and Medieval Archaeology of the University of Vienna, LBI for Archaeological Prospection and Virtual Archaeology based on an international partnership, Institute of Photogrammetry and Remote Sensing of TU-Vienna) will co-operate within the project. They have long-term experience in archaeological prospection, especially aerial archaeology and photogrammetry.

Aerial archaeology is one of the oldest archaeological prospection methods. Although it is highly efficient in detecting sites, its full potential (i.e. the detailed mapping of archaeological and palaeoenvironmental structures) can only be used, when the photographs are (ortho)- rectified, interpreted and mapped. Georeferencing of the aerial images is commonly determined with ground control points (GCPs), whose measurement and identification is a pretty time-consuming operation. It is therefore the biggest bottleneck impeding large-scale archaeological mapping projects with thousands of images.The general aim of the project therefore was to create automated solutions for georeferencing and orthophoto generation of archaeological aerial photographs (APs). The project was therefore applied for and executed from an archaeological perspective. The main focus was consequently put on creating an efficient workflow for the archaeological use of aerial photographs, which imply some special issues, as the use of both vertical and oblique images, a variety of non-metric (uncalibrated) cameras, lack of number and/or good distribution of possible ground control points and especially the use of archival (i.e. digitized analogue) material.In the scientific field of archaeology, we managed to create and compile hardware and software allowing a fast and straightforward use of APs in archaeology. In our hardware approach, we established a cost-effective hard- and software solution to record all indispensable exterior orientation parameters during image acquisition and automatically derive a footprint of the recorded AP to be used in GIS-based archives.The software solution dramatically decreases the interaction time during the process of OP generation. Although no fully automated general workflow is available, we have managed to reduce the interaction to a minimum providing a process-chain, where only a few measurements by hand are necessary to orthorectify a complete block of oblique APs.More details on the project results (including presentations and open access publications) can be found on the project website: http://arap.univie.ac.at.