Accuracy assessment of per-field classification integrating very fine spatial resolution satellite sensors imagery with topographic data

The paper describes an automated system for classifying land use on a local scale, integrating remote sensing and GIS data, with an aim of extracting meaningful information for performing spatial analysis in urban contexts. First, the background in the extraction of contextual information from fine spatial resolution satellite imagery in an integrated RS/GIS system is considered in detail. Then, a per-field classification technique is described. After that, the assessment of classification accuracy is discussed with regard to the nature of the problems emerged in the procedure. The per-field classification technique was applied to an IKONOS multispectral image with spatial resolution of 4m by utilising a digital topographic map and the classification accuracy is also reported.