Journal of Archaeological Science vol:34 issue:5 pages:830-841
Archaeological research in the territory of Sagalassos is a multidisciplinary project covering scientific disciplines traditionally linked to archaeology and also new technologies such as very high-resolution remote sensing with sufficient radiometric and spatial resolution (<2.5 m). This paper focuses on the evaluation of GIS-, pixel- and object-based techniques for automatic extraction of archaeological features from Ikonos-2 satellite imagery, which are then compared to a visual interpretation of ancient structures. The study was carried out on the unexcavated archaeological site of Hisar (southwest Turkey). Although all techniques are able to detect archaeological structures from Ikonos-2 imagery, none of them succeed in extracting features in a unique spectral class. Various landscape elements, including archaeological remains, can be automatically classified when their spectral characteristics are different. However, major difficulties arise when extracting and classifying archaeological features such as wall remnants, which are composed of the same material as the surrounding substrate. Additionally, archaeological structures do not have unique shape or colour characteristics, which can make the extraction more straightforward. In contrast to autos matic extraction methods, a simple visual interpretation performs rather well. The methods presented in this paper can be applied with variable success to archaeological structures composed of the same material as the surrounding substrate, which is often the case. (c) 2006 Elsevier Ltd. All rights reserved.