Top Secret Topographies: Recovering Two and Three-Dimensional Archaeological Information from Historic Reconnaissance Datasets Using Image-Based Modelling Techniques

摘要

Historic aerial reconnaissance photographs are a valuable resource for archaeological and other landscape-oriented research. However, they are often difficult to access and convert into useable geospatial datasets. This article outlines a workflow for producing georeferenced two and three-dimensional data products from a range of historic frame camera imagery for use in GIS environments using readily available cost-effective computer vision based techniques. These techniques are highly automated and thus more accessible to non-specialist users, and are often able to recover meaningful, useable data from imagesets which can be difficult to process using traditional photogrammetric techniques. This approach is demonstrated on a number of historical datasets which cover the Kreuttal region of Lower Austria, an archaeologically rich area which contains traces of human occupation and land-use from the Neolithic to the Modern Historic periods. WWII Allied reconnaissance aerial photographs and declassified imagery intelligence data from the US Government HEXAGON program are processed using these techniques and the resulting orthophotograph mosaics and historic digital elevation models are analysed against high-accuracy reference datasets of known quality. In addition to obviating some of the difficulties inherent in processing historic imagery, this approach also demonstrates that historic remote sensing images can be put to novel uses that far exceed their intended purposes.