For presentation at the Joint Workshop of ISPRS Working Groups 1/1, 1/3, and IV/4: Sensors and Mapping from Space 1999, September 27-30, Honover, Germany AN EXPERT SYSTEM APPROACH FOR DETECTION OF ROADS FROM REMOTE SENSING DATA

The Decision Tree Processing Expert System (DTPES) has been used to map out road distributions over an urban/rural area. A multi-source database was employed using Geographic Information System (GIS) data and remote sensing imagery. The independent data set comprised seven fields, that attempt to represent contextual, textural, geometrical characteristics of the knowledge-based data. A tabular ASCII data file was imported into decision tree environment for creation of a classification tree. In the process of decision tree analysis, the input data was recessively partitioned into mutually exclusive exhaustive subsets which define the best response variable. The resulting classification tree was used to generate generic rules for implementation of DTPES. Each rule is a representation of each node in the tree that describes a class or probability of presence of roads pixel. The spatial distribution of these pixels was mapped out to show areas with roads and their background. Techniques such as these lead to more reliable information about land use and better informed decision-making.