Key points are not available for this paper at this time.
Abstract This paper describes the characteristics of a neural network image interpretation system that is designed to extract both rural land cover and urban land use from high spatial resolution imagery (e.g., digitized aerial photography, IKONOS imagery) and/or from relatively coarse spatial and spectral resolution remote sensor data (e.g., Landsat Thematic Mapper). The system consists of modules that a) classify remote sensing imagery into different land use/land cover types, b) segment the rural land cover information into relatively homogeneous polygons in a standard GIS format, and/or c) digitize and interpret urban/suburban land use cover polygons based on their feature attribute information with the aid of a neural network.
Jensen et al. (Thu,) studied this question.