×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Close-range imaging, long-range vision

zation of Wood
ary Research 36
ee-Dimensional
;. Journal of the
Sensing 40(S),
f Aichi.(1994).
i senta chousa
f Aichi.(1999).
senta chousa

CHALLENGES AND SOLUTIONS ON THE GENERATION OF DIGITAL
SURFACE MODELS OF FACADES USING IMAGE PROCESSING TECHNIQUES
Albert Wiedemann
FPK Ingenieurgesellschaft mbH
Feurigstr. 54, D-10827 Berlin, Germany
a.wiedemann@fpk.de
Commission V, WG V/5
KEY WORDS: Architecture, Cultural Heritage, Matching, Orthoimage, Surface, DSM
ABSTRACT:
Digital Surface Models (DSM) are essential for the differential rectification. But whereas there is a lot of experiences with the
generation of Digital Surface Models of topographic terrain, only a few experiences exist in the determination of DSM of
architectural objects. This may be a result of the high demands to the DSM in architectural photogrammetry. The paper describes
approaches to determine the required data for the DSM of facades based on image matching in already oriented images and
strategies for interpolation of the DSM of faces with in-continuities in the surface and her derivation.
KURZFASSUNG:
Digitale Oberflächenmodelle (DOM) sind für die differentielle Entzerrung von entscheidender Bedeutung. Aber während es eine
Menge Erfahrungen bei der Erstellung von Oberflächenmodelle topographischer Geländeoberflächen gibt, existieren nur begrenzte
Erfahrungen bei der Erstellung von Oberflächenmodelle von Architekturobjekten. Dies kann eine Folge der harten Anforderungen
an DOM in der Architekturphotogrammetrie sein. Der Aufsatz beschreibt Ansätze um die erforderlichen Daten für DOM von
Fassaden mit Hilfe der digitalen Bildzuordnung in orientierten Bildern zu gewinnen und Strategien für die DOM-Interpolation von
Flächen mit Unstetigkeiten in der Oberfläche und in der Ableitung der Oberflächenfunktion.
1. DIGITAL SURFACE MODELS OF FACADES erosion. Both approaches also require an enormous amount of
interactive work. In only a limited number of projects resources
1.1 Requirements for Digital Surface Models as described in Dallas (Dallas et al., 1995) are available.
Some people call simple rectified images orthoimages.
Therefore some authors have developed the term true
orthoimages to describe orthoimages determined under the
knowledge of the geometric surface structure (Dequal & Lingua
1.3 Special Requirements
Some special requirements on faced DSM distinguish them

2002). In my texts and lectures only this true orthoimages are
orthoimages at all.
There are several methods to provide the required surface data
for the generation of digital orthoimages of facades. If there is a
CAD model consisting of volumes or faces with the necessary
accuracy and level of detail already existing or derived from
geodetic or photogrammetric measurements (Wiedemann
1996), it is possible to derive a Digital Surface Model (DSM) of
the facades (Fig. 2). Another approach to determine the data for
the DSM is the usage of a laser scanner (Wiedemann 2001).
Both approaches are tedious or expensive and may not be
available in all situations.
1.2 Photogrammetric Data Acquisitions
In these situations only photogrammetric approaches may come
into account. In many cases no equipment for stereo
photogrammetry is available and bundle based restitutions may
fail due to the lack of exact defined features, like on facades of
natural stones or facades suffered from a long time weather
from topographic DSM. In architecture the relation of height
differences to image acquisition distance is much bigger than in
aerial photogrammetry. Therefore the horizontal displacement
resulting from height differences is bigger than in aerial
photogrammetry. The surfaces are not continuous. The
curvature stays identical over large areas but changes abrupt.
Therefore it is necessary to mention break lines or acquire the
data with a very high density, in the ideal case the same grid
with like the raster with of the orthoimages to be derived from
the DSM.
—479—