Geoscience Reference
In-Depth Information
5- Discussion
This paper gave a comparative analysis of using five different properties to
determine the importance of individual roads. However, each property was
only considered independently. It is also possible to combine several
properties together. One problem of this combination is that the appropriate
weights are not easy to be determined. More importantly, even using the
combination strategy, there is no obvious improvement. Following is an
example to illustrate it.
In this example, the road network was generalized by implementing the
stroke-based selection approach and the importance of each stroke was
determined according to an equal weighted combination of length and
betweenness. Results show that the maximum similarity of using an equal
weighted combination falls in between the values of using individual
of the connectivity.
Maximum similarity
Properties
1:50K
1:100K
1:200K
Length
0.847
0.773
0.680
Betweenness
0.808
0.686
0.642
Length &
Betweenness
0.840
0.770
0.674
Table 3:
The maximum similarity of using an equal weighted combination
6- Conclusion
This paper reported a comparative analysis of various properties to deter-
mine the importance of individual roads for road network generalization.
These properties include one geometric property, i.e. length; three topo-
logical properties, i.e. degree, closeness and betweenness and one thematic
property, i.e. road class. Two representative selective omission approaches,
stroke-based selection and mesh density-based elimination were imple-
mented to generalize the road network. For each approach, five different
properties were respectively used to determine the importance of strokes.
The road network of Hong Kong Island was used for testing and two
