Encoding Geometric Information in Road Networks Extracted from Binary Images
The authors discuss recent progress in extracting road networks from digital imagery. We briefly review recent developments in methods for the automatic extraction of road centre line networks and propose a related algorithm aimed at encoding the geometry of road networks with line segments. Our algorithm is inspired by ‘axial lines’, which have been defined as lines of uninterrupted movement within urban streetscapes or buildings. We show that axial lines appear as ridges in isovist fields. These are formed from the maximum diametric lengths of the individual isovists, sometimes called viewsheds, which make up these fields. We present an image-processing technique for the identification of lines from ridges, discuss current strengths and weaknesses of the method, and show how it can be implemented easily and effectively.