The Automatic Morphological Description and Classification of Archaeological Monuments from Vertical Aerial Photographs.

Abstract

For most of this century, aerial photography has been an important technique for the discovery, recording and analysis of archaeological sites. For many regions there are vast archives of aerial photographs available; however, a critical problem for archaeology is the lack of time and resources available to adequately study this rich source of information by traditional methods. A problem that has more recently come to light with the advent of regional and national archaeological databases is the subjectivity inherent in the current techniques of recording and analysis. We have developed a software toolset in conjunction with archaeologists, which provides accurate, efficient and objective means to: extract and measure monuments visible in aerial photographs; provide topographic models of them; automatically classify them; and provide integration with geographical information systems (GIS). This paper describes the archaeological context of our research. It then introduces our object extraction approach, and discusses the development of an automated morphological- topographical classification scheme for earthwork monuments. This work represents a contribution to the automatic detection and extraction of low contrast, damaged objects, in scenes where significant noise and clutter are present. It also shows image processing, cluster analysis and artificial neural networks to be suitable and complementary tools for the automatic generation and implementation of an image-driven morphological classification scheme.