of 
1990. 
e sur- 
publi- 
odesy, 
arine. 
avires 
3 exi- 
  
  
DIGITAL ORTHOPHOTO USING MULTI IMAGE MATCHING 
Detlev Woytowicz 
Institut für Photogrammetrie und Bildverarbeitung 
TU Braunschweig 
Gaußstraße 22 
D-3300 Braunschweig 
Germany 
Commission V 
ABSTRACT: 
The paper describes the new possibility for the production of digital orthophotos. With the method of 
multi image matching it is possible to include many images for the computation. In this case the 
problems with covering regions are cancelled, because the information for the whole object exist. 
First results in architectural photogrammetry will be presented. 
KEY WORDS: Multi Image Matching, Digital Orthophoto 
1. INTRODUCTION 
In all fields of photogrammetry there is a 
tendency to automate the diverse procedures as 
far as possible. This tendency is supported by 
improvements in software and hardware (e.g. 
bigger capacities, shorter computing times) and 
by the development of new algorithms. 
The same holds for the production of digital 
orthophotos which aims at a maximum degree of 
automation, too. However, in the field of 
digital orthophoto the problems with covering 
regions have not been solved yet. Uneven 
surfaces and protruding objects (e.g. columns) 
still represent a problem for the exact survey 
of pieces of architecture. 
An object can only described exactly and 
completely by means of several images from 
different angles (Fig. 1-1). 
    
  
  
orthophoto 
   
orthophoto 
Figure 1-1 : Image configuration for the 
description of an object 
In this paper I will present the method of 
digital multi image matching which is capable of 
including several pictures at once in the 
computation. It thus produces a complete image 
of the object, which is then called : digital 
multi image orthophoto. 
The realization of this method is based on the 
program system PROSurf (Photogrammetric 
Reconstruction of Object SURFaces) as described 
by  /SCHNEIDER 1991/. Similar methods were 
introduced by  /HEIPKE 1991/ and  /WEISENSEE 
1991/. 
In the following, the algorithm necessary for 
the program, and the preparatory procedures will 
be presented, as well as two examples of the 
production of digital multi image orthophotos. 
2. THE FUNDAMENTAL PRINCIPLES OF THE METHOD 
This method is based on an algorithm used for 
object-based multi-image matching /EBNER ET. AL 
1987, /WROBEL 1987/, /HELEVA 1988/. The eventual 
aim of the method is the reconstruction of an 
object surface from digital images, in 
accordance with the photogrammetric imaging 
rules. In order to achieve this, it is necessary 
to construct a functional interrelation between 
the gray values of the image [g(x',y')] on the 
one hand, and the radiometrical and geometrical 
model of the object surface on the other hand. 
When a square grid in the X-Y plane is used for 
the description of the object surface, the grid 
points will generate the functions Z(X,Y) for 
the geometrical model and  G(X,Y) for the 
radiometrical model. 
The grid areas of the geometrical model and the 
radiometrical model are called surface meshes 
and surface elements respectively. Each surface 
element corresponds to a certain gray value of 
the image (Fig. 2-1). 
With the surface structure and the orientation 
parameters given, the image coordinates of the 
surface elements can now computed. After 
transforming the image