A novel method for generating 3D city models from high resolution and multi-sensor remote sensing data
Modelling urban objects and phenomena puts extremely high demands on the acquisition and analysis of the necessary data. In this context, digital airborne camera systems offer very high spatial resolutions and the simultaneous acquisition of elevation data--either through stereo-matching or by laser scanning. However, with these data the user community faces new problems in their analysis. In response, region-based classification approaches have been developed, but these are not yet fully automatic, reliable and transferable. Hence, this paper proposes a novel hybrid and multi-scale methodology which performs a classification on all segmented levels (called 'Classification on multiple Segment levels', ComS). First investigations validate the applicability as well as the accuracy of the proposed methodology compared with conventional approaches, such as a supervised, pixel-based maximum likelihood classification.
No Reference information available - sign in for access.
No Citation information available - sign in for access.
No Supplementary Data.
No Article Media
Document Type: Research Article
Affiliations: University of Osnabrück Research Centre for Geoinformatics and Remote Sensing (FZG) PO Box 1553 D-49364 Vechta Germany
Publication date: February 1, 2005