EEE ER
EEE
————
ESF
Set say - 0.1 mm, Syop - 0.04 mm (s.
tab. 4) and Sygag = 0.06 mm (i.e. 0.1% of
distance to the object) you get à Scrip-
POINT of 0.123 mm. So it is within ID.
required accuracy range.
The topics completeness of morpholo-
gical details and the fidelity of charac-
teristic shapes essentially depend on the
sampling rate and the quality of data
collection. According to the formula (s.
/Kraus 1985/) kind and size of details
are the most important factors for the
correct choice of the minimal sampling
rate (compare formula (2)). Figure 8 su-
mes up some mesh sizes due to radius of
details which are characteristic for the
natural stones of Limburg monastery.
DX -B*scnr DPOINT+PRpETAIL Q)
e
* DX(Sample) (mm)
A A
10 20 30 40 50 60
DR(Surface) (mm)
figure 8: variation of sampling rate DX
as a function of radius of details DR
For the comparison of two data sets
it is necessary to know if all topics
mentioned above are to be taken into ac-
count. It is clear that the accuracy of
the control points and the detection of
gross errors are important. Aspects refe-
ring to morphological details and fide-
lity of characteristic shapes are not so
important if the comparison is not reali-
sed using for each epoche. It means that
the differences are evaluated diretly at
the gridpoints.
trate measurement a first -
ample
Because statements about quality and
reliability of interpolated DEMs either
empirically using references or with ana-
lytical methods (compare /Reinhardt
1991/) can be made. On the other hand in
our case two data sets have two be compa-
red and the expenditure has to be as low
as possible. Therefore, the following
strategy was developed in correspondance
with the participants:
representation of an entire surface
only for visualization
direct comparison of the differences of
the gridpoints represented by single
profiles
subdivision of one surface in - if ne-
cessary - more sampling areas (compa-
rable a progressive sampling) where the
densitiy of sampling rate varies,
repetitive measuring of some profiles
per model to get an idea about the in-
dividual accuracy of the operator.
The chosen strategy has the advantage
that the accuracy of the differences de-
pend only on the variables introduced in
formula (1). Influences caused by the in-
terpolation of an DEM are avoided. In fi-
gure 9 an example for a small area of a
surface of a natural stone is depicted
(compare also fig. 2, right). The measu-
rement was carried out on an analytical
plotter P3 (Fa. ZEISS), mesh size 3 mm,
number of gridpoint = 243. The small dit-
ches are clearly visible.
figure 9: visualization of a smali part
of a surface of a natural stone contai-
ning a ditch
A further aspect for surveying of
surfaces of natural stones is given by
the fact that the stones have low
contrasts by itself. In order to support
the quality of measurement the photos
were taken with an artificial light
source (spot-light). The surface was
lighted from upper left and it was yiel-
ded an effect like hatching. This proce-
dure has the advantage that the operator
gets an very three-dimensional view in-
stead of the small differences in eleva-
tion on the surface,
