A REGION-BASED, GRAPH-THEORETIC DATA MODEL FOR THE INFERENCE OF 2ND-ORDER THEMATIC INFORMATION FROM REMOTELY-SENSED IMAGES

A graph-theoretic data model, XRAG (eXtended Relational Attributed Gra ph), is described. The model and its associated data structure can be used to represent the structural properties (morphological and symboli c) of, and relations (spatial, topological, non-topological, quantitat ive and symbolic) between, discrete regions identified in a digital re motely-sensed image. The objective in developing this model is to allo w second-order thematic information about the scene to be inferred fro m an analysis of these properties and relations. The paper briefly out lines the application of this model and an associated set of analytica l techniques to infer land Erse from an initial land cover map derived from a digital remotely-sensed image.