CIPA 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
(Pomaska, 2002). Calculation is clicked on the 
toolbar. Image files are selected one by one. First, 
single image orientations are made. Then, single 
image orientations are made. Then, two image files 
are clicked simultaneously for multi image 
orientation. This is followed by multi-image 
orientation of three, four, ... files. For each group, 
results are displayed. If there are problematic points, 
they are detected and calculation is repeated. 
6. Images are measured. This includes the definition of 
three dimensional points and polylines on the 
elevation surfaces. In other words, three dimensional 
points and three dimensional polylines that have key 
importance for the definition of the elevation surface 
are measured. This process includes the clicking of 
the same point or line on all related images in 
sequence. After the selection of a point on the first 
image, "epipolar lines" help the identification of the 
same point on the other two images. Then, compute 
command is chosen. According to the decided scale of 
the end-drawing, the limit of delta values that is to be 
accepted at the end of computation is decided on. In 
this project, 2 cm has been taken as the limit. This 
means the acceptance of measurement mistakes upto 
0.5 mm in 1 to 50 scale (Figure 4). 
7. Measured points and polylines are transformed to 
CAD environment. "DXF transfer" on the toolbar is 
clicked. A dialogue box opens. The extension of the 
source file is selected as ".POO", and that of target 
file is selected as ".DXF". "3D Interface" is checked. 
Finally, "preview and transfer" is clicked. Then, 
autocad program is turned on. The folder with DXF 
extension is opened. Three dimensional view is 
clicked under the view option. Coordinates of similar 
points such as comers of column bases are compared 
with each other to detect problems. If the values are 
very different from each other, then there is a problem 
in the transformation. It should be repeated. Since the 
east facade of Defensionskaseme is worked under 
three sub-parts, the transformation should be carried 
out for each part separately. This necessitates the 
transformation of a single part first. Then, the autocad 
file including the transferred data is saved with a 
different name. All data except the information about 
coordinate system and rotation are erased. The 
transformed data related with the second part is 
copied here. The first part is also copied. The process 
is repeated for the third part. 
Figure 4. Measurement of 3D points and polylines. 
5. PRODUCTION OF 3D MODEL 
The strategy was to visualise the monument at its present state. 
Therefore, a problem of visual reality was faced with. 3D model 
was created in autocad environment on the basis of the 
photogrammetric survey. This 3D model was processed in 
various ways so as to produce textured model, etc. 
6. IMAGE RECTIFICATION 
Image processing procedures for warping an electronic image 
on the basis of the geometric principles so that the perspective is 
controlled and an orthogonal image is achieved at the end are 
known in photogrammetry as rectifications. By the help of 
rectification, a lot of details can be mapped in a very fast way 
(Pomaska, 1998). While applying the rectification technique, 
the following points should be considered: A single tilted photo 
and four control points are enough to rectify a scaled image. But 
it is better to take photographs of building elevations as much 
perpendicular as possible to the elevations. Evaluating only the 
central part of the image also increases the accuracy of the 
result. Another important point is the organization of image 
series. Images taken from a single building elevation should 
define different planes of the elevation. In other words, a special 
rectification is necessary for every plane at different depths. It is 
impossible to rectify an object surface that is not a flat plane 
(Grussenmeyer, et all, 2002). Series of rectified images of a 
building surface can be assembled in two basic ways. These are 
ortho-imaging and texture mapping. 
Rectified images are brought together in an image processing 
software to produce ortho-images (ortho-photos). Ortho-image 
of a building elevation is a combined photo of the elevation 
without any perspective. If rectified images are assembled so as 
to cover up a simplified 3D model of a building, then a textured 
3D model is produced. Model simplification is found necessary 
since it is difficult to prepare a separate rectification for every 
building element. This simplification involves the definition of 
some basic geometric planes on the model surfaces. The 
rectification applications carried out in Defensionskaseme are 
explained in the below. 
6.1 Ortho-imaging 
In the study, Rolleimetric MSR and Photoshop are used for the 
process. The Rolleimetric MSR software package 
(http://www.rolleimetric.de) provides scaled representations of 
existing objects on the basis of rectified digital images. Images 
taken with medium format camera are used. The strategy was to 
produce an overview of the east elevation and a detail of the 
same elevation. For the overview, the elevation was abstracted 
to five basic planes whose depths varied. These planes are 
named as left comer, left wing, entrance, right wing and right 
comer. The architectural elements such as pilasters, pilaster 
bases, window sills, etc. obviously present varieties in their 
depths. But these are considered as the same for the rectification 
of the overview. Depending on these decisions, the metric 
images were reduced in size using Photoshop and saved as PSD 
files. Then, coordinates of the comers of the planes were 
introduced to MSR by clicking data-data editor-plane editor on 
the toolbar. 
Here, camera data were also entered. Then, measure - plane 
points is selected. The comers of the planes were marked one by