AN OPERATIONAL SYSTEM FOR AUTOMATED ROAD DATABASE
UPDATING FROM AERIAL IMAGERY
C. Zhang, E. Baltsavias
Institute of Geodesy and Photogrammetry, Swiss Federal Institute of Technology Zurich, ETH Hoenggerberg, CH-8093 Zurich,
Switzerland — (chunsun, manos )(a)baug.geod.ethz.ch
Commission III, WG I11/4
KEY WORDS: 3D Road reconstruction, color, context, knowledge base, edge matching, DSM/DTM; roadmarks: multiple cue
integration; performance evaluation; semi-automated system
ABSTRACT:
This paper presents a practical system for automated 3D road network reconstruction from aerial images using knowledge-based
image analysis. The system integrates processing of color image data and information from digital spatial databases, extracts and
fuses multiple object cues, takes into account context information, employs existing knowledge, rules and models, and treats cach
road subclass accordingly. The key of the system is the use of know
ofthe results, working in 2D images and 3D object space, and use o
ledge as much as possible to increase success rate and reliability
f 2D and 3D interaction when needed. Another advantage of the
developed system is that it can correctly and reliably handle problematic areas caused by shadows and occlusions. This work is part
of a project to improve and update the 1:25,000 vector maps of Sw
itzerland. The system was originally developed to process stereo
images, but it has been modified to work also with orthoimages, thus making it applicable to sensors of unknown geometry. The
system has been implemented as a stand-alone software package, and has been tested on a large number of images with different
landscape. In this paper, various parts of the developed system are discussed, and the results of our system in the tests conducted
independently by our project partner in Switzerland, and the te
presented together with the system performance evaluation.
I. INTRODUCTION
The extraction of roads from digital images has drawn
considerable attention in the last few decades. This is
increasingly stimulated by various existing and emerging
applications requiring in particular up-to-date, accurate and
sufficiently attributed road databases. The fact that most
vendors of commercial photogrammetric, remote sensing and GI
systems do not offer anything useful regarding automation of
road extraction (not even practical semi-automatic methods)
stresses the importance of this research topic. The existing
approaches for road extraction cover a wide variety of
strategies, using different resolution aerial or satellite images. À
quite extensive overview of such approaches is given in Zhang
(2003a). Semi-automatic schemes require human interaction to
provide interactively some information to control the extraction.
Roads are then extracted by profile matching (Vosselman and
Knecht, 1995), cooperative algorithms (McKeown et al, 1988)
and dynamic programming or LSB-Snakes (Grün and Li, 1997).
Automatic methods usually extract reliable hypotheses for road
segments through edge and line detection and then establish
Connections between road segments to form road networks
(Wiedemann et al. 1998). Data from different sources is often
useful (Price, 1999). Contextual information is taken into
account to guide the extraction of roads (Heipke et al., 2000).
Roads can be detected in multi-resolution images (Baumgartner
and Hinz, 2000), while Hinz and Baumgartner (2003) use
context and scale-dependent models for extraction of urban
roads in large-scale images. Several applications use map
Information (Gerke et al., 2003). The map data is used either as
approximation to start tracking or optimization process by
Snakes (Bordes et al., 1997; Agouris et al., 2001), or to search
for new roads (Vosselman and de Gunst, 1997). The road maps
can be updated by map-image matching (Klang, 1998). The
105
st results with orthoimages in a test site in the Netherlands are
existing approaches show individually that the use of road
models and varying strategies for different types of scenes are
promising. However, all the methods are based on relatively
simplistic road models, and most of them make only insufficient
use of a priori information, thus they are very sensitive to
disturbances like cars, shadows or occlusions, and do not
always provide good quality results. Furthermore, most
approaches work in single 2D images, thus neglecting valuable
information inherent in 3D processing.
In this paper, we present a knowledge-based system for
automatic extraction of 3D roads from stereo aerial images
which integrates processing of color image data and existing
digital spatial databases. The system has been developed within
the project ATOMI (Automated reconstruction of Topographic
Objects from aerial images using vectorized Map Information),
in cooperation with the Swiss Federal Office of Topography
(L*T), with aims to improve road centerlines from digitized
1:25,000 topographic maps by fitting them to the real
landscape, improving the planimetric accuracy to Im and
providing height information with 1-2m accuracy (Eidenbenz et
al., 2000). The usual input data include 1:16,000 scale color
imagery, with 30cm focal length, and 6094/2096 forward/side
overlap, scanned with 14 microns at a Zeiss SCAI. The other
input data include: a nationwide DTM with 25m grid spacing
and accuracy of 2-3/5-7m in lowland/Alps, the vectorized map
data (VEC25) of 1:25,000 scale, and the raster map with its 6
different layers. The VEC25 data have a RMS error of ca. 5-
7.5m and a maximum one of ca. 12.5m, including
generalization effects. They are topologically correct, but due to
their partly automated extraction from maps, some errors exist.
In addition, DSM data in the working area with 2m grid spacing
was generated from stereo images using MATCH-T of INPHO.
3
