The Application of Digital Techniques to the Detection and Extraction of Archaeological Earthwork Monuments from Aerial Photographs

Aerial photographs are a valuable resource for archaeological assessment, but their accurate interpretation is extremely labour intensive. The application of digital image processing techniques to the assessment process has, to-date, primarily focused on providing automatic measurement and correction of distortions inherent in oblique images. However, the monument discovery and mapping process has remained a task for the human operator. The automatic detection and mapping of archaeological evidence from aerial photographs poses significant challenges for digital image processing. The major sources of these difficulties are: • the extremely low signal-to-noise ratio of most monuments visible in aerial photographs; • the fact that earthwork monument boundaries do not adhere to any strict morphological constraints, other than the fact that many can be loosely defined as ’sub-circular’ closed loops; • the fact that nearly all monuments are visible only as thin boundary features, with no morphologically or texturally recognisable internal features; • the presence of clutter primarily modern day objects of relatively high contrast, such as walls, houses, trees, and roads; and • monument damage and occlusion, both of which further add to the problem of incomplete boundary