bul 2004 
rimposed 
. The 
2 Sensing 
:rlands. 
  
  
LARGE SCALE ORTHOPHOTOGRAPHY USING DTM FROM TERRESTRIAL LASER 
SCANNING 
A. Georgopoulos, M. Tsakiri, C. loannidis, A. Kakli 
School of Rural & Surveying Engineering, National Technical University of Athens, Greece 
Commission V 
KEY WORDS: Cultural Heritage, DEM/DTM, Orthorectification, Laser Scanning, Point Cloud 
ABSTRACT: 
The production of orthophotographs at large scales for architectural photogrammetric applications faces a number of problems. The 
main difficulty arises when the ratio of the elevation differences on the object surface to the distance from the camera is large, or 
when there are surfaces with poor definition or little texture. In these cases the standard automatic DTM production algorithms fail 
to produce a useful product. A digital surface model (DSM) from laser scanning could be used as an alternative. This paper explores 
the contribution of laser scanner data, the improvement in the accuracy and the level of automation for the production of large scale 
orthophotos. A case study is presented using data collected from a 15" century Byzantine church comprising a variety of surfaces. In 
addition to conventional geodetic and photogrammetric data acquisition, a Cyrax 2500 laser scanner was used to collect data from 
varying surfaces. Comparisons between orthophotographs from conventional procedure and combined use of photogrammetry and 
laser scanning are made to highlight the advantage of the latter in eliminating the need for lengthy photogrammetric DSM extraction 
and editing, in particular for the geometric recording of monuments and archaeological sites. 
1. INTRODUCTION 
Orthophotography is a powerful tool of aerial photogrammetry 
applied in several fields, especially after the appearance of the 
digital photogrammetric procedures. Clearly, it has the 
qualitative merits of a image document and the metric attributes 
of a map, as it is an photographic orthoprojection. However, 
orthophotography is not fully accepted by the user community 
for applications related to geometric documentation of cultural 
heritage monuments. Architects and archaeologists are reluctant 
to concede working with orthophotographs instead of the 
traditional vector line drawings. As a consequence 
orthophotography usually is not included in the standard 
specifications of the geometric recording of monuments. The 
situation is becoming worse due to the need for special 
instruction for planning and executing the photographic 
coverage to face the problems of orthophoto production for the 
monuments at large scales (i.e. 21:100). The major of such 
problems are (Mavromati et al., 2002a, 2002b and 2003): 
* Large elevation differences compared to distances between 
the camera and the object 
* Presence of “vertical” surfaces, i.e. surfaces parallel to the 
camera axis 
* Convergence of camera axes, often due to space limitations 
* Failure of automatic DTM production, as all available 
commercial algorithms are tailored to aerial images 
* Necessity for large number of stereomodels in order to 
minimize occluded areas 
* Difficulty of surveying convex objects. 
For the first two problems special measures should be taken 
during both field work and processing of the data. They are the 
main source of practically most difficulties encountered in 
producing orthophotographs and the relevant mosaics. The 
elevation. differences call for elaborate description of the 
objects surface, in order to allow for the orthophotography 
algorithm to produce accurate and reliable products. 
Usually, problems due to the image central projection and the 
relief of the object (e.g. occlusions or complex surface) can be 
solved by acquiring multiple photographs from many points of 
view. This may be compared to the true orthophoto production 
for urban areas (Baletti et al., 2003). However, processing can 
be seriously delayed for DTM generation requiring possibly 
intensive manual interaction or even a complete failure to 
produce a reliable model. 
The recent appearance of terrestrial laser scanning has already 
shown promising contribution in overcoming such problems 
(e.g. Barber et al., 2002; Bitelli et al., 2002; Drap et al., 2003; 
Guidi et al, 2002) and also confronting other similar 
applications (Baletti & Guerra, 2002). The volume of points, 
which can be over 2 million points per scan, and high sampling 
frequency of laser scanning offers a great density of spatial 
information. For this reason there is enormous potential for use 
of this technology in applications where such dense data sets 
could provide an optimal surface description for applications of 
archaeological and architectural recordings. 
Although laser scanning data may provide the surface models 
for orthophotography thus eliminating the need for lengthy 
photogrammetric surface extraction and editing, it is important 
to ensure that the resolution of a laser scan makes provision for 
the features of interest so that these features are visible in the 
resulting point cloud. Furthermore, the points in the cloud 
should be checked so those incorrectly measured due to 
multipath or mixed-pixel effects are identified and eliminated. 
Several investigations in the past have seriously considered this 
aspect and have proposed several procedures for accuracy 
assessment and specification proposal for integrating laser 
scanner data into the photogrammetric procedure, especially as 
far as the geometric recording of monuments at large scales is 
concerned (Barber ct al., 2003; Bochler et al., 2003; Lichti et al., 
2002).