ind, 
e, volume, 
edicine, or 
decade, the 
lose range 
ography at 
objects of 
he popular 
1 acgisition 
'cture and 
ents in the 
sured and 
10ds based 
ic cameras 
etry, close 
the digital 
have been 
jc. Non- 
mples, the 
ed into the 
processing 
lose range 
ion of the 
lose range 
t would be 
phy do not 
of digital 
neras. The 
cometry of 
systematic 
availability 
nds ealier 
ors. The 
uch as for 
possibility 
] introduce 
can not be 
possible or 
ELS OF 
s. 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
integrated geometric data and semantic information on 3D 
realistic models of monuments and buildings. The geometrical 
data for such information systems are determined with the 
integrated data from topographical information systems based 
on aerial photogrammetry and data on 3D.realistic models 
obtained by close range photogrammetry. The Visually 
Realistic 3D Models (VRM) of the objects, which are very 
usefull for comprehensive documentation and analysis of 
historical monuments and buildings, contain; 
e the Numerical Frame of Reference (NFR) for their 
structure, and 
e the Texture Image Database (TID) for the selected 
surfaces of the objects. 
The NFR is the simple form of 3D presentation which describes 
the main parameters of a structure. Such form is sufficient for 
planning and monioring of objects in many applications. 
However, in case of the detailed analysis of the objects, for 
example some historical monuments, the Realistic 3D Models 
are needed with all the structural details and texture. For these 
purposes, the close range digital methods, based on images 
taken from the amatour digital cameras, are becoming popular, 
because of their economical aspects. However, they require 
some precautions to prevent the suitable accuracy of the 
determined geometrical data. The most important are listed as 
follows; (1) The camera has to be checked for the level of 
stability of the interior orientation and generation of image 
systematic errors. The amatour cameras have the unknown and 
variable at time and conditions interior orientation. The images 
are taken from variable close range distances and with variable 
focusing of the camera. Therefore the camera has to be 
calibrated under required photographical conditions and 
checked in respect to its stability. (2) To reconstruct the 
numerical frame of reference for the object in the required 
reference coordinates system some control points have to be 
determined in that system by the precise surveying methods. 
These points (premarked or natural features) have to be properly 
distributed around the architectural structure and well identified 
on images. The points which describe the frame of reference 
(NFR) of the object are usually natural points but they have to 
be carefully choosen to be well identified on photographs. (3) 
The set of photographs for the whole object have to be carefully 
planned to prevent the proper geometrical accuracy of 
photogrammetric reconstruction of the object and to identify 
all points on photographs required for determination of the 
frame of reference and textural data of the object. (4) The 
suitable package for close range processing which permits to 
reconstruct accurately the shapes of all bundles of rays and 
orient them in respect to the object reference coordinate system, 
and therefore to execute the reconstruction of the frame 
reference of the object and the collection of the textural data. 
There are many examples of the real 3D modeling used for 
documentation of historical haritage. Some are presented in 
(Kolbl, Cherradi, Hostettler, 2000; Hongo, Matsuoka, 2000; 
Bartolotta, di Naro and others, 2000; Streilen, Hanke, 
Grussenmeyer, 2000). 
In the experiments carried out in the Department of 
Photogrammetry at the Warsaw University of Technology two 
packages Orient/Orpheus and PhotoModeler were used for 
reconstruction and modeling of the of the two objects; the inside 
part of the construction of the Main Aula at the Warsaw 
University of Technology (Bujakiewicz, Kowalczyk, Zawieska, 
2003) and some elements of the old building at Warsaw. For the 
first object the photographs were taken with the digital camera 
Nikon E990. During a’priori camera calibration all interior 
orientation parameters, including image systematic errors, were 
determined (Kowalczyk, 2003). It has been found that the 
interior geometry of bundles was quite stable during series of 
photos taken with the same conditions. The efficiency of the 
correction of the image systematic errors was analysed on base 
of 15 photographs taken for the 2D test. The differences in the 
mean square errors evaluated on base of 56 control points were 
within 0.2 + 0.3 pixels. The stability of the interior orientation 
was checked by taking the series of 10 photos for the 3D test. 
It has been found, that the focal distance and the principal point 
were stable within 2 + 4 pixels. In fig. 2.1, two stereo photos for 
the West part of the Aula are shown. After pre-correction for 
the image systematic errors the programme Orient/Orpheus was 
used for the reconstruction and modelling. The absolute 
orientation of the model was executed with the accuracy of 
0.016 m. The frame reference of the object with the selected 
surfaces is presented in fig. 2.2 
  
  
  
  
  
Fig. 2.2. The selected surfaces for visualization surfaces of the 
west part of the Aula. 
  
Fig. 2.3 Visualization of 3 D model with the textural data