Humanistic approaches toward automatic road extraction from raster images

Automatic road extraction has been a reemerging topic in the field of geographic information systems (GIS) in recent years because of rapidly changing landscapes and the development of temporal databases. So far, human digitizing to update old databases containing road information is still widely used. This, to some degree, demonstrates the effectiveness and reliability of human intuition in finding roads. There are six sequential algorithms developed based on the viewing mechanism and the conceptual model of a road. The results from two sample images are generated in a fast and accurate fashion. These humanistic approaches are suitable for automatic road extractions and enlarging the chance of reducing human involvement in updating road information in GIS databases.