3D-WIREFRAME MODELS AS GROUND CONTROL POINTS FOR THE AUTOMATIC EXTERIOR 
ORIENTATION 
Thomas Läbe, Karl Heiko Ellenbeck 
Institute of Photogrammetry 
University of Bonn 
laebe@ipb.uni-bonn.de 
Commission ll, Working Group 5 
KEY WORDS: Automation, Orientation, Database, Control point model, Building extraction 
ABSTRACT 
The bottleneck of today's automation of image orientation is the identification of control points for exterior orientation. A 
solution for this problem is presented. It is based on 3D-wireframe models of buildings as ground control points. The paper 
describes the setup of a database of such control points and the use of the data for an automatic exterior orientation. 
1 INTRODUCTION 
Today we are on the way towards the regular use of digi- 
tal photogrammetric workstations progressively allowing au- 
tomation of in principle all steps of image analysis. In nearly 
every case one needs the orientation parameters of the im- 
ages. Examples of actually available automatic orientation 
tasks are the interior orientation, the relative orientation and 
aerotriangulation. Regarding the different matching tasks 
of the orientation processes one finds out why solutions for 
the exterior orientation are rare. The following list gives an 
overview: 
e matching between simple 2D-objects and images cer- 
tainly is the most simple task in case the object can be 
described by its intensity image. Interior orientation for 
example can be automized by using classical template 
matching techniques. 
e matching between images and images is of similar 
simplicity due to the already identical representation. 
Quite a number of matching techniques are available 
which may be used for relative orientation or for point 
transfer in aerial triangulation. 
e matching between complex 3D-objects and images is 
used for image interpretation, e. g. for building detec- 
tion but also for the important task of control point 
identification and mensuration necessary for determin- 
ing the exterior orientation. 
Solutions which have been proposed for the automation of 
the exterior orientation are 1.) the use of signalized control 
points and 2.) the use of old image templates of natural 
control points. 
Putting the signals on the ground by hand is costly. Signals 
also have to be larger for automatic procedures than for man- 
ual measurements due to the lack of adequate identification 
procedures which can compensate for varying background. 
Different templates may be used for different scales of the 
images. Approach 2), i. e. using image templates, has to 
cope with many problems such as different illumination, veg- 
etation, scales and perspective of the images used for the 
setup of the control points and the images which have to be 
orientated. 
We propose to use 3D-control point models as a solution for 
the matching task. In most cases the problems mentioned 
218 
above do not occur. Our module for the automatic exterior 
orientation uses sets of 3D-edges as control point models. 
When one sets up a database of control point models for 
this approach, one could in principle use all 3D-objects which 
are projected in the images as straight lines. But in most 
cases one wants to use one model for many different areal 
photographs, which have been taken at different times. Be- 
cause of this, the 3D-edges have to be time invariant. Edges 
at roads or buildings are such structures. Sometimes the 
edges of roads, especially of small ones without any mark- 
ings, vary slightly because of different seasons or weather. 
Our approach, therefore, uses 3D-wireframe models of build- 
ings as ground control features, though actually only the 3D- 
edges of these wireframes are used for the orientation. Using 
wireframes eases the acquisition of the 3D-structures. Other 
possibilities are also conceivable. 
2 MOTIVATION 
The motivation for the development of a module for au- 
tomatic exterior orientation (AMOR) and the setup of a 
database of 3D-control point models was the change of the 
orthophoto production from an analytical to a digital system 
at the Landesvermessungsamt (State Survey Department) 
North-Rhine-Westfalia in Bonn. 
For nearly 30 years each map sheet of the orthophoto base 
map 1:5000 was updated at least every 5 years with a new 
orthophoto from an actual photo flight with an image scale of 
about 1:12500. With the beginning of this renewal program 
a database of time invariant control points was established to 
provide the necessary control points for each spatial resection 
at the orthophoto projector. These time invariant control 
points consist of two points on roofs (gables) of a building. 
In the area of North-Rhine-Westfalia, about 70% of the more 
than 8000 orthophoto map sheets can be oriented by a spatial 
resection using these roof points. 
For the digital production process we proposed to enhance the 
control points (2 roof points) to contol point models consist- 
ing of wireframe models of the houses or groups of houses. À 
module for automatic exterior orientation then has at first to 
detect automatically the control point models in the image, 
secondly to match the 2D image edges with the corresponding 
wireframe edges and lastly to compute the spatial resection. 
The building of the new database is at work at our Institute 
and this paper reports on this for the first time. Besides the 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996 
ho 
ho 
the 
Th 
wa 
foll