(a)
(b)
(c)
Figure 14. Image data set.
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(j)
Figure 15. (a) through (c) results of the proposed vectorization methodology on the three data sets, (d) through (f) results of the SORM on
the three data sets, and (g) through (i) results of the IEC and fuzzy ellipse-shaped clustering on the three data sets
T
able
6. C
omparison
of
the
P
roposed
M
ethod with
T
wo
O
ther
I
mage
S
pace
C
lustering
M
ethods
for
R
oad
V
ectorization with
R
espect
to
Q
uality
M
easures
Data sets
Method
RMSE
Completeness
Correctness
Junction RMSE
Ikonos
image
Proposed method
0.53
91
89
2.65
SORM
0.61
72
80
2.50
IEC & fuzzy clustering
0.54
89
86
2.11
Aerial
image
Proposed method
0.50
90
97
3.53
SORM
0.76
75
87
3.56
IEC & fuzzy clustering
0.56
81
89
2.94
QuickBird
image
Proposed method
0.38
92
93
4.20
SORM
0.67
86
89
4.07
IEC & fuzzy clustering
0.55
88
91
3.25
118
February 2016
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
