Object-oriented per-parcel land use classification of very high resolution images

The traditional method to gain information on land use in urban areas is based on the visual interpretation of very high resolution aerial photographs; a process which is both time consuming and expensive. Very high resolution satellite data are an alternative for updating and maintaining cartographic and geographic data bases. If the full potential of the new image data is to be realized for urban land use mapping, new inferential remote-sensing analysis tools need to be applied. This is because urban land use is an abstract concept which is defined in terms of function rather than form. This paper investigates the potential of an object-oriented classification approach to discriminate between different urban land use categories from panchromatic and multispectral IKONOS-2 data, covering parts of the City of Vienna. In order to validate the results, the classification was compared with a land use data set created by the City Council of Vienna (MA-41) on the basis of orthophotos.