The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part B5. Beijing 2008 
Our previous research (Borbas, 2003; Fekete, 2008) suggested 
that the two-component, transparent resin, which is in semi- 
polymerized state during its production, is suitable medium for 
the markers (in the current case, small, steel balls with 
diameters between 0.1-0.5 mm). These markers will stay at 
their positions after the end of the polymerization of the resin 
and are suitable to become the points of the test network during 
the following tests. The thickness of the plates can be regulated 
and set to constant by the quantity of the resin used during the 
production. Figure 2 shows a larger plate (3) with markers (4) 
prepared for test on a smaller casting tray (1) and a precision 
water level(2), which ensures the equal thickness of the resin 
over the whole extent of the plate. 
Figure 2. X-ray test-field 
X-ray images were taken at the Radiology Clinic of the 
Semmelweis University. Slanted images were taken of the test 
field from four directions., Network design regulations for 
multi-stage convergent photogrammetric networks were taken 
into consideration for the shooting arrangement. (Fraser, 1996; 
Fekete, 2006). For X-ray the shooting distance does not play 
such an important role as in traditional photogrammetry. The 
test field used for shooting is shown in Figure 3. 
A 
Figure 3. X-ray test-field during shoot 
To determine the image coordinates of analogue images a 
ZEISS PK-1 monocomparator was used. The object coordinates 
were calculated by a software based on Direct Linear 
Transformation algorithm and developed on our Department 
(Schrott, 2005). 
Processing accuracy analysis is enabled by the fact that the real 
location of the markers in the resin plates as well as their 
coordinates determined by photogrammetry were both known. 
The interpretation of accuracy means how close the final result 
of calculation is to the ’’real” figure. Given that we have 
sufficient number of photogrammetrically perfect object-side 
points then the square sum of coordinate differences will 
provide a global picture on the accuracy of the procedure 
applied. Accuracy was characterised as an average difference of 
distances namely the square root of the aggregate of variances 
(a). The resulted precision attributes with and without gross 
error filtering are summarized in Table 1. Figures in the table 
are in mm and mm 2 . 
Ox 2 
Gy 2 
a z 2 
(o*wwr 2) 
Non- 
filtered 
0.0143 
0.0145 
0.0153 
0.21 
gross 
error 
filtered 
0.122 
0.0119 
0.0121 
0.19 
Table 1. Accuracy analysis results of X-ray photogrammetry 
2.2 Computer Tomography 
Computer Tomography (CT) is an enhanced X-ray imaging 
technique, and it can produce 3D data very fast, that is highly 
important in case of mass data collection. The question was 
simple: is the resulting accuracy appropriate for our 
requirements? We have worked out protocols for both 
sequential and spiral CT devices to standardize the output of CT 
imaging. We designed a test to qualify the geometric accuracy 
of the CT imaging: distances of the same points were measured 
both on the skull and the resulted 3D model (Figures 4 and 5.). 
The RMS error remained within the range of 1mm both in the 
measurements of a skull and a cadaver head. Anthropologist 
experts evaluated this accuracy as sufficient for face 
reconstruction purposes. 
Figure 4. Skull in a CT device 
818