Object-based image information fusion using multisensor earth observation data over urban areas

Abstract

At present, the majority of the world's population is living in urban areas. Cities undergo constant development in their morphology. The latter is always a turned-into-stone representation of the coetaneous social, economical and technical values. The technical developments in recent years of very high-resolution spaceborne earth observation methods enable mapping of large urban areas with a decent level of detail. Additionally, detailed elevation information of urban areas in developed countries is widely available. Digital surface models (DSMs) support the classification of urban structures beyond two-dimensional classifications. We present a hierarchical, object-based and transferable framework to extract the urban structure on a high level of geometric detail for two test sites in Germany. The results show accuracies of above 90% for the land-use/land-cover classification for both test sites applying the same routines. DSMs from various sources have been utilised for the extraction of the individual building structures with accuracies of 90% and 80%, respectively. The methodology is suited to extract the urban structure on the level of individual buildings and the results can be utilised as 3D city model for the purpose of decision-making, urban planning and further analyses.