The object-oriented paradigm has also led to a new genera-
tion of databases called object-oriented (OO). Some com-
mercial OODBS are offered by the vendors of DB soft-
ware, for instance ObjectStore, Ontos, Versant, ODE, Oa,
Itasca, Objectivity /DB etc. that are realized as extensions
of C++ and Lisp.
In D. Schmidt/D. Fritsch (1994), D. Fritsch/D. Schmidt
(1995) two object-oriented DTM integrations have been
proposed: the first was implemented in the language E us-
ing the Exodus Storage Manager (SM) of the University of
Wisconsin, and the second uses the programming language
Python. Both implementations presuppose the existence of
persistent objects. The following classes were implemen-
ted:
DTM This is the highest level of hierarchy. Every access
on DTM data is handled by defined or virtual (i.e.
only declared) functions of this data type.
GRID Only height values are stored in a grid structure.
There is no topological information necessary.
TIN The data is organized by vertices, edges and faces.
A grid can easily be converted to a TIN structure.
HYBRID Data is organized as a grid with some meshes
containing TIN data structures.
GMF Geomorphological features that are subdivided in
subtypes
Heightpoint Point with z, y, z-value (i.e. peaks)
Borderline The border of a DTM, border of areas
of equal height or undefined height
Breakline Interuption in surface continuity
Structure line Height value increases or de-
creases monotonically along the line.
The implementation of these four classes is demonstrated
by the data structure of fig. 8, for instance if the DTM
class TRIANGLE is used. The class LINKS needs some point-
ers backwards, and NODES must be notified to store three-
dimensional coordinates and two-dimensional ones. 'The
low-level classes NODE and LINK have counters to store the
number of references from each class in the mid-level hier-
archy. Every class can be accessed by a spatial index tree
(R-Tree).
WEE WEST si gmp Fo Mh
Cn TIN 1 (oa wor Object dem)
Eu 2 AE
©,
Repessived Part of 2 "o.
Figure 8: 2.5D data structure of an object-oriented integ-
rated TIN
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
6 CONCLUSIONS
The integration of height information into GIS databases
is an important topic for present and future R&D in this
field. Although some progress could be reached in the re-
cent past, the solutions offered so far are not yet fully sat-
isfying.
With the evolution of object-oriented techniques new but
probably simpler techniques of height data integration are
feasible. New standards will evolve and hopefully merge
soon to make the programming effort more valuable by
using a common interface.
It seems reasonable to have first a 2.5D data integration.
This boundary description is especially desired in topo-
graphic mapping. The link of 3D objects with a 2.5D de-
scription comes out with a complete modelling of real world
objects that is restricted on being stationary. The data
display can be solved by public domain software products,
for instance Geomview. This is an interactive program for
viewing and manipulating geometric objects. It allows in-
teractive control over the point of view, speed of movement,
appearance of surfaces and lines (see fig. 9)
Figure 9: Geomview control panel with grid D'TM and at-
tached buildings in camera view
In figure 9 Geomview is used as display engine for 3D spa-
tial data. Geometric descriptions will be sent object by
object with an internal name used to reference the geo-
metry. The coordinates will then be send back to the main
program together with the name of the selected object. In
this way, points or objects can be selected for processing
in more complex queries. It is also possible, to dynamic-
ally change the geometry, color etc. of the object or adding
other objects.
This last visualization gives a concluding impression on
the possibilities of efficient height data integration. The
real world is three-dimensional — it does not matter how
this 3D world is mapped into spatial databases. The main
point is the availability of 2.5D and 3D capability in GIS
what should be realized in an efficient and robust manner.
Bode, T.,
periences
geologicall
Informatio
35-44.
Bonham-C
tems for C
Breunig, |
gischer Ol
Gesellsche
Hannover
Chen, X.,
ating Dele
Archives
nich, Ger
Chen, X.,
of GIS bi
Internatio
Sensing, !
Fritsch, L
digitale C
sion, Reil
Fritsch, |
three-dim
in geogr:
Archives
nich, Ger
Fritsch, 1
in GIS. In
Hobbie,
Houlding
Berlin, 3
INTERC
scription
Kasper,
mension:
Analyse
GIS in |
Stuttgar
Kufoniyi
logy as
