PRACTICABLE PHOTOGRAMMETRY FOR 3D-GIS 
Klaus Tempfli, Morakot Pilouk 
International Institute for Aerospace Survey and Earth Sciences (ITC) 
P.O.Box 6, 7500 AA Enschede, The Netherlands 
Tel. +31-53-4874358, Fax. +31-53-4874335 
E-mail: TEMPFLI@ITC.NL, MORAKOT@ITC.NL 
XMII ISPRS Congress, Inter-Commission Working Group III/IV 
KEY WORDS: GIS, CAD, Modelling, Mapping, Object Reconstruction, Visualization 
ABSTRACT 
Encouragement for 3D-GIS comes from advances in technology and evolving user needs to cope with increasingly complex 
spatial analysis tasks. The demand for spatial information is most pressing in urban areas. The present paper elaborates on 
how current photogrammetric means can be used for 3D topographic mapping for municipal information systems. Aiming 
at comprehensive spatial 3D operation we rely on a data model that is based on 3D-FDS. Beside the basic issue of adequate 
data modelling and providing tools for analysis and visualization, efficient data acquisition is in fact the major challenge from 
the point of view of costs of creating a 3D-GIS. Even without automated information extraction from images, aerial and 
terrestrial photogrammetry offer solutions to both geometric object reconstruction and texture mapping. For testing the 
applicability of concepts and available tools we have built a prototype 3D-GIS. We also report our experiences with regard 
to the developed procedures and make suggestions for further development of photogrammetric means. 
1. INTRODUCTION 
3D-GIS is receiving increasing interest in R&D circles with 
possible applications in disparate fields. Being concerned 
here with the possible contribution of photogrammetry to 
3D-GIS, our interest is not on spatial modelling of sub- 
surface phenomena nor on global environmental change, 
but exclusively on topographic objects. These are objects 
with discernible boundaries such as buildings, streets, man 
holes, the ground surface--objects traditionally subject to 
surface mapping. Three-dimensional (3D) refers to three 
spatial reference dimensions and not to 2D plus a time 
dimension. 
The demand for spatial information is most pressing in 
urban areas. In the USA 7596 of the population already lives 
in urban areas. It is predicted that at the turn of the 
millennium almost 50% of the world's population will live in 
cities. Urban planning and management of such enormous 
conglomerates becomes a rapidly growing problem. 
Prevailing initiatives to satisfy the demands for spatial 
information are the establishment of 3D-CAD models and 
2D-GISs; the first supporting (architectural) design issues and 
the latter, municipal management. Designing, manipulating 
and graphically presenting 3D objects can conveniently be 
done by commercially available CAD systems. On the other 
hand, inventories, the analysis of spatial relationships and 
association with a multitude of thematic properties, are 
available in GIS. 3D-CAD models, also referred to as 3D city 
models, have already been built and used on a project basis 
(see, eg, Rinner, 1993; Grün et al, 1993; Gruber et al, 1995; 
859 
Ranzinger et Lorber, 1995). They offer photo realistic views, 
animation, and interactive 'walk through' when linked up 
with virtual reality browsers. They do not supply analytic 
analysis of relationships between objects. Even current GISs 
fall short in fumishing 3D topology, and visualization of 
aspects of an urban scene by 2D line maps is unsatisfactory 
in many cases. 
A 3D-GIS that is also capable of feeding 3D-CAD models 
can overcome the above indicated problems. To this end 
we need a data model that can cope with geometry, 
topology, and semantics. Chapter 2 identifies requirements 
for an urban 3D-GIS and describes the spatial data model 
based on the formal data structure (FDS; Molenaar, 1992). 
One of the big challenges in 3D-GIS is efficient data 
acquisition. À great deal of information about objects with 
discernible boundaries can be extracted from images, 
specifically from stereoscopic images when 3D objects are 
the matter of concern. By employing photogrammetry we 
can take full advantage of its 3D measuring capabilities as 
well as the possibility of image mapping for photo true 
visualization. In chapter 3 we describe procedures readily 
applicable by building on commercially available 
photogrammetric tools. Since our interest is directed at 
production-oriented environments, we shall not review 
progress made in (semi-Jautomated building and road 
extraction, but will elaborate on manual (point) 
photogrammetry, considering both analytical and digital 
plotters. Leaving object recognition to the human 
interpreter, we concentrate on object reconstruction. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996