Annett Faber
DETECTION OF DOMINANT ORTHOGONAL ROAD STRUCTURES IN SMALL SCALE
IMAGERY
Annett Faber, Wolfgang Förstner
University of Bonn, Germany
Institute of Photogrammetry
annett | wf@ipb.uni-bonn.de
KEY WORDS: Coding of orientations, MOMS-02, Road detection, Segmentation.
ABSTRACT
The objective of the presented work is the automatic segmentation of urban areas from high resolution satellite images,
such as MOMS-02 images or from aerial images taken from high altitude flights. The structure of urban areas, as seen
from satellites or aeroplanes, is mainly characterized by three elements: the road network, the morphology of the built up
areas and the distribution of the vegetation. There exist many types of road structures in large cities, which govern the
local topology and geometry of the individual roads. Typical examples are orthogonal networks, star type networks or
irregular networks. Seen world wide, orthogonal networks appear to be the most common ones, as e. g. to be found in
Mannheim, Barcelona, New York or Canberra. The paper presents an approach for segmentation of dominant orthogonal
road structures from high resolution satellite images, like MOMS-02, or aerial images.
1 INTRODUCTION
The structure of urban areas, as seen from satellites or aeroplanes, is mainly characterized by three elements: the road
network, the morphology of the built up areas and the distribution of the vegetation. The road network immediately
attracts one's special attention for its geometrical arrangement which reveals its man made origin. The morphology of the
building areas can be described in two levels: one is complementary to the road network and shows the overall structure of
the settlement, the other is the local arrangement of the buildings within building blocks. Road network and building areas
are in a particular way injected by vegetation. Roads may be partially or totally bounded by trees, e. g. when forming
allees. Built up areas show irregularly structured vegetation, especially in gardens. Moreover, parks or cemeteries often
form larger closed areas within cities, either at the boundary of a town or completely surrounded by built up areas. Though
all three components are closely related we start with the analysis of the road structure as it appears to be recovered most
easily and gives the most essential information about the overall city structure. Due to their diversity in structure and in
appearance vegetation and built up areas are less significant for identifying city structures.
2 MOTIVATION
There exist many types of road structures in large cities, which govern the local topology and geometry of the individual
roads. Typical examples are orthogonal networks, star type networks or irregular networks. Seen world wide, orthogonal
networks appear to be the most common ones, as e. g. to be found in Mannheim, Barcelona, New York or Canberra.
Hyppodamos called the grid network the "Rationality civilized behavior" [(Sennet, 1991), S. 70].
There exist early works on the extraction the road structure from low resolution images, but mainly of rural road (Barzohar
and Cooper, 1996, Fischler et al., 1981, Fischler and Heller, 1998). These approaches ignore the intersections of roads and
so they lose the topology of the road network. In contrast to it (Baumgartner, 1998, Hinz et al., 1999, Steger et al., 1995,
Steger et al., 1997) focus on the generation of descriptions of the road grid with topology intact, but only focus on rural
roads too. Price (Price, 1999) is the only one he extract road networks in cities and with consideration the topology. For
modeling he needs to interactively provide three points that give the location, direction and spacings of the road network.
This step could be replaced by an automatic method to find dominant directions, as we will present it in the following
sections.
Our task is the automatic segmentation of urban areas from high resolution satellite images, like MOMS-02 images, based
on road structures. The segmentation should lead to a partitioning of the city area into regions with a dominant orientation
of the roads within a region and different orientation of the roads in neighboring regions. Such a segmentation can then
be used to ease the extraction of individual road segments in each region, as their orientation is known. Moreover, roads
274 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.