Martin Smith 
  
ABSOLUTE AND EXTERIOR ORIENTATION USING LINEAR FEATURES 
Martin J. SMITH, David W. G. PARK 
Institute of Engineering Surveying and Geodesy 
The University of Nottingham, UK 
martin.smith @nottingham.ac.uk 
david.park @nottingham.ac.uk 
Working Group IC/11 
KEY WORDS: Orientation, photogrammetry, algorithms, geo-referencing, image registration, image mapping. 
ABSTRACT 
For many years we have had digital photogrammetric systems available that can perform automatic interior and relative 
orientation of aerial photographs. Developments in absolute and exterior orientation are not so robust because the 
problem is significantly more difficult. The reason for the increase in difficulty is that there is no easy way to create the 
master image or target patch which represents the traditional ground control point. If we have such an image then there 
are a number of techniques that are well established to undertake area or feature based matching to find the 
corresponding point on the photograph. There are also an increasing number of users of photogrammetric techniques 
and software who would like to be able to provide control for absolute or exterior orientation in a very easy and efficient 
way without having to rely on careful identification of ground control points. These two issues have been the motivation 
to investigate the general issue of improving the flexibility and ease of providing ground control for the processes of 
absolute and exterior orientation using analytical and digital techniques. 
This paper will present a new approach to absolute and exterior orientation based on the measurement of linear features 
and their automatic matching. The results show that the technique can produce results suitable for many environmental 
mapping applications and in some instances is suitable for topographic mapping. It is also shown to have the capability 
of automatic relative orientation. 
1 INTRODUCTION 
1.1 Digital Photogrammetry 
The advances in digital photogrammetry have brought many changes to traditional photogrammetric procedures and 
products. Although these advances are bringing exciting changes to the users of photogrammetry there is still a role for 
analytical photogrammetry to play, particularly in traditional vector mapping and in many special engineering 
applications. Perhaps the most important benefit brought by digital photogrammetry is the potential for automation. 
Having the photographs in digital image format provides them in a suitable form for computer manipulation and 
processing. Automation is important for a number of reasons although the main drive is the quest to bring about 
productivity gains by reducing the time taken to perform a task. From the point of view of job satisfaction there are 
mixed feelings, as automation can remove the monotony and tedium of a repetitive task. For example digital elevation 
observation can de-skill a task such as performing relative orientation or leave the photogrammetrist to clear up the 
problems of unsatisfactory matching in digital elevation modelling (DEM) generation during quality control. There are 
some tasks where automation can bring consistency of observation and quality. This might be noticeable when a human 
Observer tires and for example may result in a reduction in their ability to perform relative orientation. 
One of the side effects of introducing automation and de-skilling the tasks is that it enables a wider group of people to 
use the techniques, largely as tools. Many traditionally accepted photogrammetric techniques are being included in 
standard remote sensing image processing software. These can now be seen and used as tools by a non-specialist 
photogrammetric community. It is easy to see that this will increase in the future. This ‘non-specialist’ group might 
include geologists and environmental managers, it might also include archeologists, geographers, planners and users of 
remote sensing. To these users, the highest metric accuracy is often not of major concern as the nature of the detail 
/information being extracted is not accurately definable, it is typically ‘soft’ detail. What is important to these users is 
that the tools must be easy to use, with minimal ‘specialized’ skill required. 
  
850 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 
  
tre 
"-— DÀ OD Cu 0 Oo oc 
€) t MC CF pd 
—