Performance evaluation of data fusion techniques for archaeological prospection based on satellite data

Abstract

Listen

Translate article

The application of Very High Resolution (VHR) satellite imagery to archaeological prospection can furnish useful information for the identification of archaeological features, related to ancient land use patterns, irrigation networks, paleo-hydrological systems, roads, walls and buildings. These archaeological features could be enhanced by using data fusion techniques which are able to merge the complementary characteristics of panchromatic and multispectral images. The quantitative evaluation of the quality of the fused images is one the most crucial aspects in the context of data fusion. This issue is particularly relevant in the case of the identification of archaeological features, because data fusion could enhance or lose the small spatial and spectral details which are generally linked with the presence of buried archaeological remains. This study is focused on the evaluation of data fusion algorithms applied to Quickbird images for the enhancement of archaeological features. Three different data fusion techniques, Gram-Schimdt, PCA, and wavelet, were applied to a study case located in the South of Italy. Focusing on the archaeological features, the evaluation process was performed by using two different protocols with and without a reference image. Results obtained from the two protocols showed that the best performance was obtained from the wavelet data fusion.