identified control points which were measured on image in 
mono mode. The stereo approach, applied in photogrammetric 
digital workstations, allows to increase the orientation accuracy 
of satellite images, by including in this process some stereo 
measured tie points (of any type including poorly identified) in 
addition to a very few control points with known field location 
(Hole, 2011). When the suitable number of tie points is 
uniformly distributed around 3D photogrammetric model then 
the required control points (surveyed in field) would be 
significantly reduced. In such case not only the accuracy of 
exterior orientation of imagery is increased but also a load of 
field work is much decreased. The most important advantage of 
using a photogrammetric digital workstation is a possibility to 
generate a 3D model and measure all specific spatial vector data 
for each building within the area, in process of stereo- 
digitization. When the photogrammetric System, such as 
Summit Evolution, is on-line integrated with 3D GIS platform, 
such as ArcGIS (ArcMap module), then all measured feature 
data are directly transferred to the selected layers of GIS 
database and supplemented with the attributes acquired from 
photographs or some other sources [Bujakiewicz, Preuss, 2010]. 
Such data subsequently can serve for 3D modelling of buildings 
and visualization. Direct connection of the measuring 
photogrammetric system with 3D data base allows also to 
superimpose all available data which are archived in this base, 
such as ground footprints from the cadastral files or from 
orthophoto, to the photogrammetric model generated from 
VHRS images. This is helpful not only for stereo-digitization of 
the new 3D vector product but also to control and update the 
existing data. However, in any stereo photogrammetric system, 
the accuracy of measurement and recognition and 
discrimination of object’ features depend on type and resolution 
of source images. As it was investigated in some earlier 
research, concerning processing of data from VHRS images, the 
accuracy of features measurement can be assumed due to 
standard of LoD2, but recognition and discrimination of some 
features qualify such 3D models rather to level LoD]. 
3. PRESENTATION AND ANALYSIS OF RESULTS 
In the experiment, two stereo IKONOS-2 panchromatic images 
of one of the Cracow city districts were used. They were 
acquired by the Satellite Center of Regional Operations (SCOR) 
in Komorow, Poland in May 2005, with time interval of 1 
minute. The stereo images used in this experiment had been 
reprocessed by the cubic convolution resampling, due to spatial 
and radiometric resolution. The images were delivered with 
RPC coefficients and file of metadata with all detailed 
information. The part of Cracow area selected for 
photogrammetric stereo compilation is shown in figure 2. 
  
Fig. 2 Area of Cracow city selected for compilation 
(maps.google.pl) 
To support exterior orientation and for the final accuracy 
evaluation, some well identified man-made control and check 
points were surveyed in field. The example of such type of 
point is shown in figure 3. 
   
Fig. 3 Image location of one of control point 
Other data available for the purpose of this project were: 
terrain elevation data - DTED Level 2 (in NATO standard 
format) and digital orthophoto from midle scale photography, 
which was used for mono plotting of building basements and 
consequently served as the reference data. 
Process of exterior orientation of two Ikonos images and their 
stereo compilation were carried out on the photogrammetric 
system Summit Evolution combined with ArcGIS, which 
accepts the Ikonos imagery and their corresponding RPC 
coefficients. The wide range of the Summit Evolution projects 
(depending on various input data) is shown in the figure 4. 
  
  
Select the kind of project: 
  
     
in 
  
A SEES 
Aerial Project ADS40 Digital SocetSet and SAR Stereo Orthophoto 
Sensor Leica ADS40 Feature Collection 
a OD 
Close Range DPPDB PCI Epipolar Space Imaging 
Model IKONOS RPC 
   
  
  
  
Cancel | 
Fig. 4 Window of Summit Evolution showing the possible 
input data (Manual of Summit Evolution) 
  
As it is obvious, the exterior orientation stage has a major 
influence on accuracy of 3D model reconstruction and all data 
determined and therefore a great effort went to this part. For 
evaluation of the exterior orientation accuracy in relation to a 
number of tie and control points, different combinations of 
points were selected. From the principle, stereo measurement of 
tie points reduces the number of control points with known 
ground coordinates and therefore allows improvement of the 
exterior orientation accuracy by inclusion of more observation 
equations without the additional cost for field load. The 
adjustment process uses two types of observation equations 
which are based on both coplanarity and collinearity conditions. 
In total, 22 field surveyed points, uniformly distributed 
around the model area, had XYZ coordinates (in PUWG-92 
system), of which, 5 were used as the control points and the 
remaining 17 as check points. For accuracy analysis, The RMS 
was evaluated for check points (CHP), after exterior orientation 
of stereo images of Ikonos and. reconstruction of 3D model in 
Summit Evolution photogrammetric workstation, using of RPC 
114 
th