Effects of image characteristics on the identification and extraction of archaeological features from Ikonos-2 and Quickbird-2 imagery: case study Sagalassos (southwest Turkey)

This paper compares and evaluates the significance of spectral characteristics and pixel resolution of the very high spatial resolution satellite systems (VHSRS) Quickbird-2 and Ikonos-2 for automatic extraction of ancient features on one archaeological site, the antique town Sagalassos (southwest-Turkey). The evaluation of the spectral characteristics is based on a band-by-band comparison. Pixel-and object-based classification techniques are compared with visual analyses to assess the effects of different image characteristics. The analysis reveals that Quickbird-2 is outperforming Ikonos-2 for the visual identification of ancient remains. Its augmented spatial resolution however does not necessarily result in a better automatic classification. For Sagalassos, maximum likelihood classification on Ikonos-2 gives the best results. The accuracy of automatic extraction depends on the type and characteristics of the VHSRS data, the classification method and on on-site parameters. Site characteristics appear to play a limited role in object-based extraction on Quickbird-2, but become very important on Ikonos-2.