air.
ling points
1 both im-
raints like
lirectional
1sions can
the corre-
match us-
rbergen et
one image
other im-
1g the cor-
pyramidal
ge dispar-
ity search
Observed
ge.
ute image
tified im-
ach cam-
hieved by
3D model
ned in an
nce link-
il., 1998).
| baseline
nse depth
olution is
iewpoints
implified
lotion re-
the nec-
| models.
mesh to
del to the
systems.
riangular
| a corre-
triangles
1e corre-
s texture
idence 1s
structed.
r discon-
th maps
The texture itself can also be enhanced through the multi-
view linking scheme. A median or robust mean of the
corresponding texture values can be computed to discard
imaging artifacts like sensor noise, specular reflections and
highlights (Koch et al., 1998, Ofek et al., 1997).
To reconstruct more complex shapes it is necessary to com-
bine multiple depth maps. Since all depth-maps are located
in a single metric frame, registration is not an issue. To
integrate the multiple depth maps into a single surface rep-
resentation, the volumetric technique proposed in (Curless
and Levoy, 1996) is used.
3 RESULTS
In this section a few results are presented to illustrate the
possibilities of our approach.
A first example is one of the Dionysus statues found in
Sagalassos. This 2m high statue was placed on the mon-
umental fountain on the upper market-square of Sagalas-
sos. The statue is now located in the garden of the mu-
seum in Burdur. In our case it was simple to record a 1-
minute video. Bringing in more advanced equipment such
as laser range scanner would much more difficult. In Fig-
ure 5 different steps of the reconstruction process are il-
lustrated. The 3D model was obtained from a single depth
map. A more complete and accurate model could be ob-
tained by combining multiple depth maps (see further). A
more smooth look could be obtained for the shaded model
by filtering the 3D mesh in accordance with the standard
deviations that are obtained as a by product depth compu-
tation. This is not so important when the model is textured
mapped with the original images.
A second example is shown in Figure 6. It is a Medusa
head which is located on the entablature of a monumental
fountain in Sagalassos. The head itself is about 30cm large.
The 3D model was obtained from a short video sequence.
In this case also a single depth map was used to reconstruct
the 3D model. Notice that realistic views can be rendered
from viewpoints that are very different from the original
viewpoint.
An important advantage of our approach compared to more
interactive techniques (Debevec et al., 1996, PhotoMod-
eler) is that much more complex objects can be dealt with.
Compared to non-image based techniques we have the im-
portant advantage that surface texture is directly extracted
from the images. This does not only result in a much higher
degree of realism, but'is also important for the authenticity
of the reconstruction. Therefore the reconstructions ob-
tained with this system can also be used as a scale model
on which measurements can be carried out or as a tool
for planning restorations. A disadvantage of our approach
(and more in general of most image-based approaches) is
that our technique can not directly capture the photometric
properties of an object, but only the combination of these
with lighting. It is therefore not possible to re-render the
3D model under different lighting. This is a topic of future
research.
Figure 5: 3D reconstruction of Dionysus. (a) one of the
original video frames, (b) corresponding depth map, (c)
shaded view of the 3D reconstruction, (d) view of the tex-
tured 3D model.
(a) | (b)
(c) 0
Figure 6: 3D reconstruction of a Medusa head. (a) one
of the original video frames, (b) corresponding depth map,
(c) and (d) two views of the 3D model.
—583—
