Geoscience Reference
In-Depth Information
ure 3
. The value of similarity increases slowly. After passing a peak, the
value of similarity begins to decrease sharply. This is because the mesh
density-based elimination approach implements an opposite processing
compared with the stroke-based selection approach.
Furthermore, the distributions are quite similar in despite of using different
properties.
It is because that, in each elimination, only several strokes (i.e.
those bounding strokes of a mesh) are involved Tto be ranked. Thus it has
more probabilities for these strokes to have the same ranking in despite of
using different properties.
Table 2
lists the maximum similarity of using
four properties.
Maximum similarity
Properties
1:50K
1:100K
1:200K
Length
0.871
0.779
0.742
Degree
0.870
0.782
0.759
Closeness
0.873
0.767
0.747
Betweenness
0.870
0.780
0.738
Table 2
:
The maximum similarity of using four properties
Results in
Table 2
show that the similarities of using four properties are
very similar. Using length does not have obvious advantages any more. It
may be because the dead end roads which do not belong to a boundary of
any mesh are not involved in calculation of the similarity. And using
degree property seems to perform a little better.
4.2.2 Evaluation of connectivity
As stated by Chen and his collaborators (2009), “a mesh is a loop, when
one or more of its segments are omitted, the remainder becomes part of a
new larger loop formed by one or one originally adjoining meshes”. There-
fore, when the mesh density-based elimination approach is implemented,
the connectivity of the retained network can be preserved in despite of
using different properties.
