bul 2004
inces on
/els are
st like
oftware,
> city of
est city
deration,
] by single
ialized by
le) at the
his level is
navigation
] by single
ction are
ed by two
direction.
detail as
nal level.
'sented in
onal levels
or higher
he most
ematical
ere done
cy were
utomatic
ematical
hen they
anguage.
nue. For
by using
x))y) A
tartPoint
onnected
e road y.
1 there is
point can
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
After formalizing the representations, generalization operators
that will be used during the process were determined and object
representation types defined in reasoning processes. Following
figure briefly illustrates the reasoning processes in this work.
Although only displacement, merging and refinement
considered in the Figure 2, smoothing, simplification are
significant operators, which are used in line generalization.
MRDB
n ‘4th Level
R,D
M
3rd Level
MRDB
Junction
2nd Level
MRDB
Junction
Ist Level
Geometric
R : Refinement [7] Object €} Representatior
in MRDB
D : Displacement Relations
C Reasoning
M : Merging «t» Bidirectional Processes
Connectivities
Figure 2: Reasoning Processes (Modified from Kilpelainen,
1997)
4.4 Tests for the Representations
Single and two-lane representations of the junctions were tested
to determine which representation level is more consistent to
the current software in which navigational calculations can be
done. In this context network topology is created in three
different programs than same queries are done for different
representations of the junctions in the name of finding shortest
path. One of the test results is given in Figure 3 According to
this figure, arrows on the single lane representation show the
road directions. While query result points the wrong way for
this representation because of wrong directionality, it is
possible to obtain the correct result in two-lane
representation(Ulugtekin et al., 2004). As a result of these tests,
it is understood that although two-lane representation of the
roads have some problems, it is more consistent to the
navigation programs and their current solutions for the
problems.
5. CONCULUSION AND FUTURE WORKS
Navigation systems take more places in automobile industry
and everyday life day by day. Future works on intelligent
transportation systems aiming to manage the full-automated
drive, so future systems should be designed based on artificial
intelligence and expert systems. It is obvious that a user, driver,
wants the system to represent the world reality as detailed as
possible. But it should not be forgotten that the lack of detail
makes the map unreadable. However, limits of the details
should be determined by considering constraints of small
display cartography.
In this study it is proposed that using two-lane representation of
roads in navigation maps would be more convenient for several
reasons. First of all it represents the world reality, especially
junctions, better than single one. At the end of the formalization
process it is understood that there is a significant differences
between formalization of two lane and single lane
representations because of the one or two-way distinction in
single lane representation. Finally, while standard topological
algorithms can run on two-lane representations additional
programs are needed for single representations. This means that
exchangeability of the two-lane representation is better than the
single one.
Start Point 1
e = —
i End Point
End Point
Single Lane Representation
Figure 3. Shortest path query results in two and single lane
representations
In this work, actual scales of the representations were not
explicitly determined, because only road network were
considered as a problem. Thus roads should be considered with
their surrounding objects. This work should be developed and
applied on the all road network of Istanbul. Derivation of these
