ıt of data used to
voxel will then be
which data has to
ints will occur, if a
ased on the quality
atching process.
serve as weigthing
attribute values will
rve as additional
cation steps taking
purposes. Typical
Z for the averaged
ction of the image
ic qualities, the last
ch object point and
visulizations.
attribute values
asOy10715-—— IMAGE;
2:0 2,072. image;
a0y3,073; image,
and generation of
data
is used to build up
bject points (cf fig.
d by a Delauney
As result of such a
lich have to be
cted. To all of them
triangles which are
npleted by the edge
riangle surrounds a
surface. Taking all
irface is complete.
en points with their
ves are stored and
ation process the
grids is presented.
sreo models just in
1996
front of the busts face. Alltogether these grids consist of surface of a volume object. The unique description of the
about 49.000 matched points.
surface is first of all useful for visual application. A typical
visualization in this context is volume rendering, giving a
pseudo realistic view of the surface (cf fig.10). The
2
5
SEHK
VEN
SESE
: SST
SN JU) SR XX
Q QM
il
S DAD
UT AU Oo oio
NG
SS PSEA
Qi xX 1007
HH d 5
nu GS 7
2
XA
X S QUY
Se
j i^ ARE ih %
IA 0;
/ 0
QUA Se T /,
ih
i
(/
OU
h Ü
0
% VHS
Bs A
SN HR
KK KON
KT TO
Q SIN)
SON I POP
GS
ul lily Wi
A
Qu)
d
Figure 9: Perspective view of the front part of the object
The accumulation was then done with a voxel side length
of 1 mm reducing the data set to 41.000 remaining points.
Fig.9 shows a perspective view of the compiled face
constructed from the accumulated point set.
In order to get an impression of the accuracy for the
matched process the mean Z-differences within identical
areas have been calculated: 07 = 0.4 [mm]. The value of oz
is influenced by the accuracy of the orientation
parameters and the matching process and documents
here, that
> the simultaneous photo triangulation has given
reliable orientation values
P 2
EL
SORES
SNS
CO XR
CS
"^, 2
Es
RE
(A
(OA QS RRS
n ho KK SES
XXX?
is
li i i lt OS
li TOO SI
CON)
J) Wi VAS i i
J
Ü
contents of fig. 10 presents a views from an artificial
space position, what might be extended to an animation
of the object, turning the view point all around the volume.
Another type of visualization is the presentation of surface
parts with respect to a selected reference frame. As
shown in fig. 5 any individual coordiante system (eg Xs,Zs
à ) may be taken as reference. The data of the DOM then
RN has to be transformed into this frame followed by a
AL
AX]
Qe . . : ;
SU selection of all those points with have to be considered
CL LE ROSES 1 1
Se for the graphical presentation.
RAA) CST
NN
KK KIN
" ESS
ESI
A I
CSSS $8
ss
KK NR
GEXA?
SR ROSSO
KS
ASS
Figure 10: Shading of the volume from an artificial
perspective
— the low contrast has not reduced the matching quality,
giving good correspondance within the overlapping ^ This process allows a very flexible graphical investigation
areas of adjacent model areas.
of any desired part of the surface (cf fig. 11).
Figure 11 shows two graphics of surface parts as they
were calculated in the local model coordinate system.
Use of object data. The investigation does not aim at a This gives an unusual impression due to the orientation of
complete and universal GIS solution but tries to give a the coordinate systems, what is changed if the data is
procedure allowing for the establishment of a consistent transformed into a local system better matching the
data set for a DOM providing a unique description of the surface.
left model
|
/ )N
Mo (SEE TRON
(Eu)
ASS
== x =
selected plane right model
Figure 11: Graphical presentation of a surface part.
75
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
