ISPRS Commission III, Vol.34, Part 3A ,Photogrammetric Computer Vision“, Graz, 2002
METROPOGIS: A SEMI-AUTOMATIC CITY DOCUMENTATION SYSTEM
Andreas Klaus* Joachim Bauer*
Konrad Karner* Konrad Schindler!
* VRVis Research Center for Virtual Reality and Visualization
email: {klaus,bauer,karner}@vrvis.at
! Institute for Computer Graphics and Vision, Graz University of Technology
email: {schindl}@icg.tu-graz.ac.at
KEY WORDS: GIS, Image Orientation, Image Sequences, Vanishing Points, Bundle Block Adjustment.
ABSTRACT
In this paper we report on a new system to enhance a 3D block model of a real city with terrestrial measured data
of the facades. The terrestrial data consists of photographs taken by a hand-held digital consumer camera using short
baselines. The relative orientation of the photographs is calculated automatically and fitted towards the 3D block model
with minimized human input using vanishing points. The introduced city documentation system delivers a fully 3D
geographic information data set and is called MetropoGIS.
1 INTRODUCTION
Virtual city models attract attention due to their usability
for various applications such as the simulation of wind,
flooding, traffic, and radio transmission as well as city plan-
ning, surveying, virtual tourism, cultural heritage docu-
mentation, and emergency training. All these applications
have different requirements on the data set which have to
be fulfilled. Some of these requirements are:
e 3D geometric model
e high geometric accuracy
e high geometric resolution
e high resolution texture for all objects for photo-realistic
visualization
topological description of building features like win-
dows and doors.
In our system we attach importance to all these above men-
tioned requirements. In this paper we concentrate on the
first part of the whole city modeling process, the orienta-
tion of the input images. In Section 2 we give an overview
of related work. Our work flow starting with line and van-
ishing point extraction, the relative orientation of image
sequences and the mapping of the images to the 3D block
model are given in Section 3. Some results which illus-
trate the robust search for corresponding points to estimate
the relative orientation between image pairs and the image
to 3D model fitting using the vanishing points are shown
in Section 4. Section 5 presents the conclusions and gives
some ideas on future work.
2 RELATED WORK
A working system for calibrated, terrestrial image acqui-
sition in urban areas can be found in Teller [10]. The
system works with uncalibrated images and provide cali-
brated, geo-referenced images as output. To enable a fully
Figure 1: Overview of the virtual 3D block model of the
inner city of Graz.
automatic processing a GPS-based position estimation is
required. The sensor provide omni-directional images for
better pose estimation. Additional information can be found
in Antone [1], [2] and Bosse [7]. Our approach is quite
similar but differs in some important aspects. The most
relevant are that we can handle occlusions in a much easier
way, because we are evaluating multiple images and there-
fore the baselines of adjacent images are much smaller.
In addition we are much more flexible with our hand-held
camera, that can be used as well in stop and go as in dy-
namic mode.
3 OVERVIEW OF OUR WORK FLOW
In our approach we concentrate on the refinement of the
facades of buildings. We assume that a 3D block model
with roof lines exists. For our pilot project, the inner city
of Graz, we started with a simple three dimensional (3D)
A - 187
