T
E
L
KO
M
NI
K
A
,
V
ol
.
14,
N
o.
3,
S
ept
em
ber
20
16,
pp.
9
8
1~
9
86
I
S
S
N
:
1
693
-
6
930
,
ac
c
r
edi
t
ed
A
b
y
D
IK
T
I,
D
e
c
r
e
e
N
o
:
58/
D
I
K
T
I
/
K
ep/
2013
D
O
I
:
10.
12928/
T
E
LK
O
M
N
I
K
A
.
v
1
4
i
3
.
2757
98
1
R
ec
ei
v
ed
M
a
y 2
,
2
01
6
;
R
e
v
i
s
ed
J
une
2
1
,
20
1
6
;
A
c
c
e
pt
ed
Ju
ly
7
,
201
6
Fuz
z
y
C
-
M
e
a
ns
C
lu
st
er
in
g
B
a
s
e
d o
n
I
mp
r
o
v
e
d
M
ar
ked
W
a
te
r
s
he
d
T
r
a
n
s
for
m
a
ti
on
C
u
ijie
Z
h
a
o
*
1
,
2
,
H
o
n
g
d
o
n
g
Z
h
a
o
1
,
W
ei
Y
ao
3
1
S
c
ho
ol
of
E
l
e
c
t
r
on
i
c
and
I
nf
or
m
at
i
on E
n
gi
ne
er
i
n
g,
H
eb
ei
U
n
i
v
er
s
i
t
y
of
T
ec
h
nol
ogy
,
T
ia
n
jin
3
004
01
,
T
ia
n
jin
,
C
h
i
na
2
T
i
anj
i
n U
ni
v
er
s
i
t
y
of
F
i
nan
c
e
and E
c
ono
m
i
c
s
P
ear
l
R
i
v
er
C
o
l
l
eg
e,
T
ia
n
jin
3
018
11,
T
ia
n
jin
,
C
h
i
na
3
S
c
h
oo
l
of
E
co
n
o
m
i
cs
an
d
M
anage
m
ent
,
T
i
anj
i
n
U
ni
v
er
s
i
t
y
of
S
ci
e
n
ce
and
T
ec
hno
l
ogy
,
T
i
anj
i
n 3
002
22
,
T
ia
n
jin
,
C
h
i
na
*
C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
m
ai
l
:
3687
2763@
qq.
c
om
A
b
st
r
act
C
ur
r
ent
l
y
,
t
he f
u
z
z
y
c
-
m
eans
al
gor
i
t
hm
pl
ay
s
a c
er
t
ai
n r
o
l
e i
n
r
em
ot
e s
en
s
i
n
g i
m
age c
l
a
s
s
i
f
i
c
at
i
on
.
H
ow
ev
er
,
i
t
i
s
ea
s
y
t
o f
al
l
i
nt
o
l
oc
a
l
o
pt
i
m
al
s
o
l
ut
i
on,
w
h
i
ch
lea
d
s
to
p
oor
c
l
a
s
s
i
f
i
c
at
i
on.
I
n
o
r
der
t
o
i
m
pr
o
v
e
t
he ac
c
ur
a
c
y
of
c
l
as
s
i
f
i
c
at
i
on,
t
hi
s
pa
per
,
b
as
e
d on t
he
i
m
pr
ov
e
d
ma
r
k
e
d
w
at
er
s
h
ed s
e
gm
ent
at
i
on,
pu
t
s
f
or
w
ar
d a f
u
z
z
y
c
-
m
eans
c
l
u
s
t
er
i
ng
opt
i
m
i
z
at
i
on al
gor
i
t
hm
.
B
ec
au
s
e t
he w
at
er
s
h
ed s
e
gm
ent
at
i
on an
d
f
u
z
zy
c
-
m
eans
c
l
u
s
t
er
i
ng
ar
e
s
en
s
i
t
i
v
e
t
o
t
he
noi
s
e
of
t
he
i
m
age,
t
h
i
s
pap
er
u
s
e
s
t
h
e a
dapt
i
v
e
m
edi
a
n
f
i
l
t
er
i
ng
al
go
r
i
t
hm
t
o
el
i
m
i
nat
e
t
he no
i
s
e i
nf
or
m
at
i
o
n.
D
ur
i
ng
t
hi
s
pr
o
c
es
s
,
t
he c
l
a
s
s
i
f
i
c
at
i
o
n num
ber
s
an
d
i
ni
t
i
al
c
l
u
s
t
er
c
ent
er
s
of
f
u
z
z
y
c
-
m
eans
ar
e de
t
er
m
i
ne
d by
t
h
e r
es
u
l
t
of
t
he f
uz
z
y
s
i
m
i
l
ar
r
el
at
i
on
c
l
u
s
t
er
i
ng.
T
hr
oug
h a s
er
i
es
of
c
om
par
at
i
v
e s
i
m
u
l
at
i
on e
x
per
i
m
e
nt
s
,
t
he
r
es
ul
t
s
s
h
ow
t
hat
t
he m
et
hod
pr
opo
s
ed i
n t
hi
s
paper
i
s
m
or
e
ac
c
ur
a
t
e t
han t
h
e I
S
O
D
A
T
A
m
et
hod,
and
i
t
i
s
a f
ea
s
i
b
l
e t
r
ai
n
i
ng m
et
h
od.
Ke
y
w
o
rd
s
:
adapt
i
v
e
m
edi
an
f
i
lt
e
r
in
g
,
m
ar
k
ed
w
at
er
s
he
d
s
egm
e
n
t
at
i
on,
f
u
z
zy
s
im
ila
r
it
y
r
el
at
i
on
,
f
u
zzy
C
-
M
eans
c
l
us
t
e
r
in
g
C
o
p
y
r
i
g
h
t
©
20
16 U
n
i
ver
si
t
a
s A
h
mad
D
ah
l
an
.
A
l
l
r
i
g
h
t
s r
eser
ved
.
1
.
I
n
tr
o
d
u
c
ti
o
n
A
t
pr
es
ent
,
r
em
ot
e s
ens
i
n
g t
ec
h
nol
og
y
has
b
een
gr
adua
l
l
y
de
v
e
l
op
ed,
t
h
e d
e
t
ec
t
ed
i
m
age r
es
ol
ut
i
on c
a
n ac
hi
e
v
e d
ec
i
m
et
er
c
ount
i
ng.
I
n r
em
ot
e s
ens
i
ng m
oni
t
or
i
n
g
and g
eogr
a
ph
i
c
i
nf
or
m
at
i
on
ac
qui
s
i
t
i
o
n,
r
em
ot
e
s
ens
i
ng
t
ec
hn
ol
og
y
c
an
qu
i
c
k
l
y
an
d
ac
c
ur
at
el
y
o
bt
a
i
n
t
he
r
el
e
v
ant
d
at
a.
P
eop
l
e'
s
da
i
l
y
w
or
k
an
d
l
i
f
e c
an
not
b
e s
epar
at
ed f
r
om
t
he
v
as
t
am
ount
of
r
em
ot
e
s
ens
i
ng
dat
a.
H
o
w
ev
er
,
ho
w
t
o
pr
oc
es
s
m
or
e s
c
i
ent
i
f
i
c
l
y
a
nd
ac
c
ur
at
el
y
needs
f
ur
t
her
di
s
c
us
s
i
on.
W
i
t
h
t
he
i
nc
r
ea
s
i
ng
r
es
o
l
ut
i
on
of
r
em
ot
e
s
ens
i
n
g
da
t
a,
t
he
f
oc
us
of
t
he
r
es
ear
c
h
i
s
t
he
k
i
nd
of
i
m
age
pr
oc
es
s
i
ng
t
ec
hn
i
qu
es
t
o
ex
t
r
ac
t
m
or
e
ac
c
ur
at
e
qua
nt
i
t
at
i
v
e
i
n
f
or
m
at
i
on
and
m
u
lt
i
-
s
c
al
e i
nf
or
m
at
i
on.
I
n
t
he
ex
t
r
ac
t
i
on
of
r
em
ot
e s
ens
i
ng
i
m
age d
at
a,
t
h
e
bas
i
c
i
s
i
m
age
c
la
s
s
if
ic
a
t
io
n
[
1]
.
B
ec
aus
e
t
he r
em
ot
e s
ens
i
ng
i
m
age c
ont
a
i
ns
r
i
c
h
s
pec
t
r
al
i
nf
or
m
at
i
o
n,
and
t
he
dat
a
i
s
v
er
y
l
ar
g
e,
w
h
i
c
h
l
e
ads
t
o
poor
ac
c
ur
ac
y
of
c
l
a
s
s
i
f
i
c
at
i
on of
r
em
ot
e s
ens
i
n
g i
m
age.
T
her
e ar
e t
w
o m
ai
n
w
a
y
s
t
o c
l
as
s
i
f
y
r
em
ot
e s
ens
i
n
g i
m
age,
w
h
i
c
h
ar
e s
up
er
v
i
s
e
d
c
l
as
s
i
f
i
c
at
i
on
and
non
-
s
up
e
r
v
i
s
ed
c
l
as
s
i
f
i
c
at
i
o
n.
N
o
n
-
s
uper
v
i
s
ed
c
l
as
s
i
f
i
c
at
i
o
n
ha
s
bec
om
e
one
of
t
he
m
ai
n m
et
hods
i
n t
h
e f
i
el
d of
r
em
ot
e s
ens
i
ng i
m
age c
l
as
s
i
f
i
c
at
i
on
[
2]
.
F
u
z
z
y
c
l
us
t
er
i
ng
ana
l
y
s
i
s
i
s
o
ne of
t
he m
ai
n t
ec
hn
i
qu
es
f
or
uns
uper
v
i
s
ed m
ac
hi
ne l
ear
n
i
ng
.
I
t
u
s
es
t
he f
uz
z
y
t
heor
y
t
o
an
al
y
z
e
t
h
e
i
m
por
t
ant
da
t
a,
a
nd
es
t
a
bl
i
s
hes
unc
e
r
t
ai
n
des
c
r
i
pt
i
o
ns
f
or
eac
h
s
am
pl
e.
I
t
c
an
obj
ec
t
i
v
e
l
y
r
ef
l
ec
t
t
h
e
r
eal
w
or
l
d,
and
i
t
has
i
m
por
t
ant
t
h
eor
et
i
c
al
and
pr
ac
t
i
c
al
v
al
ue.
W
i
t
h
t
he f
ur
t
her
de
v
e
l
opm
ent
o
f
t
he ap
pl
i
c
at
i
on,
t
he
r
es
e
ar
c
h of
t
he f
u
z
z
y
c
l
us
t
er
i
n
g a
l
gor
i
t
hm
i
s
c
ons
t
ant
l
y
enr
i
c
hed.
A
m
o
ng t
he
num
er
ous
f
uz
z
y
c
l
us
t
er
i
ng
a
l
gor
i
t
hm
s
,
t
he f
uz
z
y
c
-
m
eans
c
l
us
t
er
i
n
g
i
s
t
he
m
os
t
w
i
del
y
us
ed
and
s
uc
c
es
s
f
u
l
a
l
gor
i
t
hm
.
I
t
c
an
obt
ai
n
t
he
s
am
pl
e
m
e
m
ber
s
hi
p
b
y
opt
i
m
i
z
i
ng
t
he
obj
ec
t
i
v
e
f
unc
t
i
on,
and
det
er
m
i
ne
a
s
er
i
es
of
s
a
m
pl
e
da
t
a,
f
i
na
l
l
y
ac
hi
e
v
e t
h
e
goa
l
of
au
t
om
at
i
c
c
l
as
s
i
f
i
c
at
i
on of
t
h
e s
am
pl
e
dat
a.
F
u
z
z
y
c
l
us
t
er
i
n
g opt
i
m
i
z
a
t
i
on ga
i
n t
he at
t
ent
i
o
n of
m
a
n
y
i
n
t
er
na
t
i
o
na
l
s
c
hol
ar
s
i
n
r
el
at
e
d
f
i
el
ds
r
api
dl
y
s
i
nc
e i
t
s
ad
v
e
nt
.
I
n 20
10
Y
and
i
Z
ar
ne
gar
ni
a a
nd H
um
i
d A
l
av
i
Maj
d
put
f
or
w
ar
d t
he
c
lu
s
t
e
r
in
g
a
lg
or
i
t
hm
bas
es
on
s
i
m
i
l
ar
m
at
r
i
x
[
3
]
.
I
t
c
oul
d
i
de
nt
i
f
y
t
he
i
nc
l
u
de
d
pr
ot
e
i
ns
i
n
es
opha
gus
,
s
t
om
ac
h
and
c
ol
on
c
anc
er
s
b
as
ed
on
s
i
m
i
l
ar
i
t
y
of
G
ene
O
nt
o
l
o
g
y
anno
t
at
i
on,
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
81
–
9
86
982
s
i
m
i
l
ar
degr
ee
bet
w
een
t
he
pi
x
el
s
c
o
ul
d
be c
a
l
c
ul
at
e.
B
ut
t
he a
l
g
or
i
t
hm
c
annot
b
e app
l
i
ed t
o t
h
e
r
em
ot
e
s
ens
i
ng
i
m
ages
,
be
c
aus
e
of
t
he
l
ar
g
e
am
ount
of
c
al
c
ul
at
i
on.
T
hen,
i
n
2
01
1
B
e
l
i
ak
ov
G
,
J
a
m
s
S
i
nt
r
oduc
ed t
he f
u
z
z
y
c
-
m
eans
c
l
us
t
er
i
ng a
l
gor
i
t
hm
,
i
t
i
s
one of
d
y
n
a
m
i
c
c
l
us
t
er
i
ng
al
g
or
i
t
hm
t
hat
m
i
ni
m
i
z
e t
h
e er
r
or
s
u
m
o
f
s
quar
es
of
s
a
m
pl
es
an
d t
he c
l
us
t
er
i
ng c
ent
er
of
a
d
y
n
am
i
c
c
l
us
t
er
i
n
g al
g
or
i
t
h
m
[
4
]
.
I
t
dec
i
des
t
h
e s
am
pl
es
be
l
on
gi
ng
t
o
w
h
i
c
h
c
at
e
gor
y
t
y
p
e
[
5
],
t
he
c
l
us
t
er
i
n
g
a
l
gor
i
t
hm
c
annot
f
ul
l
y
an
al
y
z
e
t
h
e
gr
a
y
s
c
al
e
c
har
ac
t
er
i
s
t
i
c
s
of
t
he
s
am
pl
es
and
t
he
c
onn
ec
t
i
o
n
de
gr
ee
of
a
dj
ac
ent
pi
xe
l
s
[
6
,
7
]
,
but
t
h
e
f
uz
z
y
c
-
m
eans
c
l
us
t
er
i
n
g
al
gor
i
t
hm
w
as
ef
f
ec
t
i
v
e t
o
noi
s
e
i
nf
or
m
at
i
on.
T
hi
s
paper
i
nt
r
o
duc
es
t
he
t
heor
y
of
f
uz
z
y
c
-
m
ean c
l
us
t
er
i
ng a
l
gor
i
t
hm
,
and des
c
r
i
bes
t
he
m
ar
k
ed
w
at
er
s
he
d
s
egm
ent
at
i
on,
t
hen
e
l
a
bor
at
es
t
he
i
m
pr
ov
ed
al
gor
i
t
h
m
.
I
t
needs
t
o
c
ons
t
r
uc
t
t
he
f
u
z
z
y
s
i
m
i
l
a
r
i
t
y
m
at
r
i
x
.
T
he
r
em
ot
e
s
ens
i
n
g
i
m
age
c
on
t
ai
ns
m
an
y
k
i
nds
of
obj
ec
t
i
v
e i
nf
or
m
at
i
on.
i
t
n
e
eds
t
o f
i
nd t
he p
i
x
el
of
ea
c
h obj
ec
t
,
and t
h
en det
er
m
i
nes
t
he i
ni
t
i
al
c
l
us
t
er
c
ent
er
of
t
he
al
g
or
i
t
hm
.
T
hr
ough c
om
par
at
i
v
e ex
per
i
m
en
t
s
,
t
hi
s
pa
p
er
s
how
s
t
h
e
f
eas
i
bi
l
i
t
y
an
d
s
uper
i
or
i
t
y
o
f
t
he
f
uz
z
y
c
-
m
ean
c
l
us
t
er
i
ng
opt
i
m
i
z
at
i
on
a
l
gor
i
t
hm
.
I
t
c
an
be
s
een
t
hat
t
h
e
i
m
pr
ov
ed
f
u
z
z
y
c
-
m
eans
c
l
us
t
er
i
ng
al
gor
i
t
hm
c
annot
on
l
y
i
m
pr
ov
e
t
h
e
c
la
s
s
if
ic
a
t
io
n
ac
c
ur
ac
y
,
but
al
s
o
i
m
pr
ov
e t
he
ab
i
l
i
t
y
t
o
a
v
o
i
d
l
oc
al
ex
t
r
em
u
m
.
2.
I
m
p
r
o
v
e
d
M
a
r
k
ed
W
at
e
r
sh
ed
S
e
g
m
e
n
ta
ti
o
n
A
n
d
F
u
zzy
S
i
m
ila
r
R
el
at
i
o
n
2.
1
.
I
m
p
r
o
v
e
d
M
ak
ed
W
at
er
ed
S
e
g
m
e
n
ta
ti
o
n
A
l
g
o
r
i
th
m
A
l
l
l
oc
al
m
i
ni
m
u
m
v
al
ue
of
i
m
age
c
or
r
es
pond
i
ng
t
o
t
he
s
egm
ent
at
i
o
n
r
eg
i
o
n
i
s
gi
v
en
b
y
t
he
i
m
pr
ov
ed
m
ar
k
ed
w
at
er
ed
s
egm
ent
at
i
on
al
gor
i
t
hm
,
t
he
l
oc
a
l
m
i
ni
m
u
m
v
al
u
e
no
t
o
nl
y
c
or
r
es
ponds
t
o
t
he
m
i
ni
m
u
m
o
f
r
eal
i
m
ages
but
al
s
o
t
he
ps
e
udo
l
oc
a
l
m
i
ni
m
u
m
v
al
u
e c
aus
e
d b
y
t
he
t
ex
t
ur
e
det
ai
l
s
and
b
ac
k
gr
ound
noi
s
e.
T
he
i
m
age
i
s
s
pl
i
t
t
o
t
ens
of
t
hous
a
nd
s
of
s
m
al
l
ar
ea
f
or
t
he
ps
eu
do
l
oc
al
m
i
ni
m
um
v
al
ue,
t
he
di
v
i
de
d
ar
e
a
s
s
er
i
ous
l
y
i
nf
l
uenc
e
on
t
h
e
ex
t
r
ac
t
i
o
n
of
i
m
age t
ar
get
.
I
m
age pr
et
r
eat
m
ent
m
et
hods
ar
e appl
i
ed t
o
i
m
age.
Mor
pho
l
og
y
m
i
ni
m
u
m
c
al
i
br
at
i
on t
ec
hn
ol
og
y
t
hat
bas
ed o
n m
or
phol
og
y
an
d f
u
z
z
y
d
i
s
t
a
nc
e t
r
ans
f
or
m
at
i
o
n
i
s
pr
opos
e
d
t
o
r
educ
e
t
he
num
ber
of
f
al
s
e
l
oc
a
l
m
i
ni
m
u
m
poi
nt
an
d
r
es
t
r
ai
n
o
v
er
-
s
eg
m
ent
at
i
on.
T
he
ex
t
r
ac
t
i
on a
l
gor
i
t
hm
bas
ed on m
ar
k
er
-
bas
ed w
a
t
er
s
he
d i
s
pr
opos
ed,
t
h
e al
g
or
i
t
h
m
c
an ex
t
r
ac
t
t
he
l
o
w
f
r
equenc
y
p
ar
t
of
t
h
e gr
a
di
e
nt
i
m
age,
a
nd t
ag
t
he
l
oc
al
m
i
ni
m
u
m
v
al
ue
,
a
n
d i
t
f
or
c
es
t
he
l
oc
al
m
i
ni
m
u
m
v
al
ue of
t
he
or
i
gi
nal
i
m
age
w
i
t
h
m
or
phol
o
g
y
m
i
ni
m
u
m
c
al
i
br
at
i
o
n
t
ec
hno
l
og
y
t
o
s
hi
el
d
t
h
e
or
i
gi
nal
m
i
ni
m
a
i
n
t
he
or
i
g
i
n
al
gr
adi
ent
i
m
age
[
8
]
.
T
he
al
gor
i
t
hm
al
t
ho
ugh
ef
f
ec
t
i
v
e
l
y
s
ol
v
e
t
h
e
o
v
er
-
s
egm
ent
at
i
o
n
pr
ob
l
em
,
b
u
t
i
t
m
a
k
e
t
he
edge
pr
of
i
l
e
or
t
he
d
et
a
i
l
s
of
t
he
i
m
ages
pos
i
t
i
on
i
ng
not
ac
c
ur
at
e
l
y
.
I
n or
d
er
t
o
av
oi
d
i
m
age
c
ont
our
f
u
z
z
y
or
i
en
t
at
i
o
n de
v
i
at
i
on
and
s
ol
v
e
t
h
e o
v
er
-
s
egm
ent
at
i
on
pr
obl
em
,
f
i
r
s
t
l
y
,
t
he
c
ol
or
i
m
age
m
us
t
be pr
oc
es
s
by
ad
apt
i
v
e
m
e
di
a
n
f
i
l
t
er
.
T
he
pr
oc
es
s
el
i
m
i
nat
es
t
h
e
i
m
pul
s
e
no
i
s
e a
nd
pr
es
er
v
es
t
he
det
a
i
l
of
t
h
e
i
m
age [
9
].
T
he adap
t
i
v
e
m
edi
an f
i
l
t
er
i
nc
l
u
des
t
w
o l
a
y
er
s
.
1
mi
n
2
ma
x
=
−
=
−
m
e
d
m
e
d
AZ
Z
AZ
Z
(
1)
1
mi
n
2
mi
n
=
−
=
−
x
y
x
y
BZ
Z
BZ
Z
(
2)
A
m
ong
w
hi
c
h,
”
Z
m
e
d
”
i
s
t
he
m
edi
an
of
i
m
age,
”Z
mi
n
”
i
s
t
he
m
i
ni
m
u
m
of
i
m
age,
”Z
m
a
x
”
is
t
he
m
ax
i
m
u
m
o
f
i
m
age,
”
Z
x
y
”
is
t
h
e v
a
l
u
e of
pi
x
el
.
I
f
A
1
>
0,
A
2
>0
,
t
he s
ec
o
nd l
a
y
er
ent
er
s
t
h
e
ac
t
i
v
e s
t
at
e
,
a
nd
t
he s
i
z
e
of
f
i
l
t
er
w
i
n
do
w
i
nc
r
eas
es
.
I
f
B
1
>0
,
B
2
>0
,
”Z
x
y
”
and
”Z
e
d
”
ar
e o
ut
p
ut
va
l
u
e
s
.
S
ec
ond
l
y
,
t
he gr
a
di
e
nt
i
m
age i
s
deal
t
w
i
t
h t
he ope
n and c
l
os
e o
per
at
i
ons
,
an
d i
t
i
s
r
ec
ons
t
r
uc
t
ed b
y
m
or
phol
ogi
c
a
l
.
Mor
pho
l
og
i
c
al
ope
n and c
l
os
e o
per
at
i
o
ns
bas
ed o
n
t
he
ex
pans
i
on
of
t
he m
or
ph
ol
o
g
y
an
d c
or
r
os
i
o
n.
T
h
e def
i
n
i
t
i
on
of
m
or
phol
og
i
c
al
ex
pans
i
on
oper
at
i
o
n i
s
s
ho
w
n
in
t
he e
quat
i
on
be
l
o
w
:
(
,)
m
i
n
(
,)
=
⊕
nb
D
f
r
f
nb
r
(
3)
T
he def
i
ni
t
i
on
of
t
he m
or
phol
o
gi
c
a
l
ex
pa
ns
i
o
n c
or
r
os
i
o
n i
s
s
ho
w
n
in
t
h
e e
quat
i
o
n
bel
o
w
:
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
F
u
z
zy
C
-
M
ea
ns
C
l
us
t
er
i
n
g
B
as
ed
on
I
mpr
ov
ed
M
ar
k
e
d
W
at
er
s
hed…
(
C
u
ij
ie
Z
h
a
o
)
983
(
,)
m
a
x
(
,)
=
Θ
nb
E
f
r
f
nb
r
(
4)
T
he def
i
ni
t
i
on of
m
or
phol
ogi
c
a
l
c
l
os
e
and
ope
n op
er
at
i
o
ns
f
or
pr
oc
es
s
i
ng gr
adi
ent
i
m
age i
n t
hi
s
p
aper
ar
e
s
ho
w
n
in
t
h
e eq
uat
i
on
be
l
o
w
:
(
,)
(
,)
(
,)
(
,)
=
•
=
nb
nb
nb
nb
nb
nb
C
f
r
E
f
nb
r
O
C
r
D
C
nb
r
(
5)
“
nb”
i
s
t
he
di
s
c
s
t
r
uc
t
ur
e
el
em
ent
,
“r”
i
s
t
he
r
adi
us
of
t
he
s
t
r
uc
t
ur
e
el
em
ent
.
T
he l
at
es
t
gr
adi
ent
i
m
age
i
s
ex
pr
es
s
e
d
b
y
”I
c
”
,
t
h
en
t
he
r
egi
ona
l
m
i
ni
m
a
ar
e
ex
t
r
ac
t
ed
f
r
om
t
he
m
odi
f
i
ed
gr
adi
ent
i
m
age,
t
he
ex
t
r
ac
t
ed m
i
ni
m
a c
ons
t
i
t
ut
e t
he
bi
nar
y
m
ar
k
er
i
m
age
ma
r
k
C
I
.
T
he m
ar
k
er
s
ar
e
t
he m
i
ni
m
a of
t
he or
i
g
i
n
al
i
m
age b
y
H
-
m
i
ni
m
a t
ec
hnol
og
y
.
H
-
m
i
ni
m
a el
i
m
i
nat
es
t
h
e m
ar
k
er
r
egi
onal
m
i
ni
m
a w
hi
c
h i
s
l
o
w
er
t
ha
n t
hr
es
h
ol
d
v
a
l
ue
”
h”
,
t
he
bi
n
ar
y
m
ar
k
er
i
m
age w
as
r
epr
es
ent
ed
b
y
ma
r
k
C
I
.
(|
)
=
ma
r
k
ma
r
k
C
CC
I
IMMIN
I
I
(
6)
I
MMI
N
(
)
i
s
t
h
e c
al
i
br
at
i
on
oper
at
i
o
n of
m
or
phol
o
g
y
m
i
ni
m
a.
T
he
f
o
llo
w
in
g
f
ig
u
r
e
s
ar
e
r
es
ul
t
s
of
w
at
er
s
he
d s
egm
ent
at
i
on
.
T
hr
ough t
h
e f
ol
l
o
w
i
ng
c
om
par
at
i
v
e
s
i
m
ul
at
i
on
ex
per
i
m
ent
s
,
i
t
c
a
n b
e s
e
en
t
he
i
m
pr
ov
e
d
m
ar
k
e
r
bas
ed
w
at
er
s
hed
a
l
gor
i
t
hm
i
s
m
o
r
e
ac
c
ur
at
e
t
han
t
h
e
pr
ev
i
ous
a
l
gor
i
t
hm
.
F
or
c
o
m
pl
ex
i
m
ages
,
t
he m
ar
k
ed w
at
er
s
hed a
l
g
or
i
t
hm
s
egm
ent
at
i
on
r
es
ul
t
s
ar
e
not
i
d
ea
l
.
t
hes
e c
om
pl
ex
i
m
ages
r
ef
er
t
o
t
he
i
nt
er
n
al
n
oi
s
e
di
s
t
ur
banc
e
and
m
et
i
c
ul
ous
c
l
os
e
i
r
r
egu
l
ar
c
o
m
pl
ex
i
m
age,
and
i
m
ages
w
i
t
h
l
i
t
t
l
e
d
i
f
f
e
r
enc
e b
et
w
e
en
t
ar
ge
t
a
nd
bac
k
gr
ound.
T
he r
e
as
on
i
s
t
hat
t
he
l
o
w
pa
s
s
f
i
l
t
er
i
ng a
l
g
or
i
t
hm
c
an
f
i
l
t
er
t
h
e i
m
age e
dge
i
nf
or
m
at
i
o
n
w
i
t
h s
m
al
l
gr
adi
ent
a
m
pl
i
t
ude
i
n t
he
i
m
age.
T
he i
m
pr
ov
ed
al
g
or
i
t
hm
not
on
l
y
s
o
l
v
e a
bo
v
e
p
r
obl
em
s
,
but
al
s
o
ef
f
ec
t
i
v
el
y
pr
es
er
v
e
t
h
e
edge
det
a
i
l
s
of
t
he i
m
age.
T
he i
m
pr
ov
ed m
ar
k
er
bas
ed
w
at
er
s
hed
a
l
g
or
i
t
hm
s
egm
ent
at
i
on
r
es
ul
t
s
of
i
s
m
or
e ac
c
ur
at
e,
and
i
t
i
s
a f
eas
i
b
l
e
t
r
ai
ni
n
g
m
et
hod
.
F
i
gur
e
1.
T
he or
i
gi
nal
i
m
age
F
i
gur
e 2.
W
at
er
s
hed
s
egm
ent
at
i
on
F
i
gur
e
3.
I
m
pr
ov
ed
m
ak
e
r
w
at
er
ed
s
egm
ent
at
i
on
2.
2
.
F
u
z
z
y
C
-
M
ean
s
B
a
sed
o
n
F
u
zzy
S
i
m
ila
r
R
el
at
i
o
n
T
he f
uz
z
y
c
-
m
eans
al
gor
i
t
h
m
i
s
hi
gh s
ens
i
t
i
v
i
t
y
t
o t
he i
ni
t
i
al
c
l
us
t
er
i
n
g c
ent
er
.
I
n o
r
der
t
o
av
o
i
d
t
he
i
nc
or
r
ec
t
c
l
us
t
er
i
ng
ef
f
ec
t
c
aus
ed
b
y
w
r
on
g
s
el
ec
t
i
ng
i
ni
t
i
a
l
c
l
us
t
er
c
e
nt
er
s
,
t
h
e
i
ni
t
i
a
l
c
l
us
t
er
i
n
g c
ent
er
i
s
d
et
er
m
i
ned
b
y
f
u
z
z
y
s
i
m
i
l
ar
i
t
y
r
e
l
at
i
on i
n t
h
i
s
pa
per
.
F
u
z
z
y
s
im
ila
r
m
a
t
r
ix
“
R”
i
s
es
t
abl
i
s
hed
b
y
i
nt
r
od
uc
i
ng
s
i
m
i
l
ar
r
el
at
i
o
ns
hi
p
“r
ij
”
,
“r
ij
”
r
epr
es
ent
s
t
he
s
i
m
i
l
ar
i
t
y
d
e
gr
ee
bet
w
een
“x
i
”
and
“x
j
”
,
t
he
f
or
m
of
f
uz
z
y
s
i
m
i
l
ar
m
at
r
i
x
i
s
s
ho
w
n
w
i
t
h t
he equ
at
i
on be
l
o
w
:
22
11
12
1
21
2
12
...
...
...
...
...
...
...
=
n
r
n
n
n
nn
rr
r
rr
R
rr
r
(7
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
81
–
9
86
984
“r
ij
”
is
t
he n
or
m
al
i
z
ed
r
es
ul
t
of
E
uc
l
i
dea
n d
i
s
t
anc
e f
or
m
edi
an
v
a
l
ue.
2
1
1
1
()
/
ma
x
(
)
=
=
−−
=
∑
n
ij
i
j
k
ij
ij
r
xx
n
R
r
r
(
8)
T
he
c
l
as
s
i
f
i
c
at
i
o
n
n
um
ber
s
and
t
he
s
am
pl
es
of
ea
c
h
c
at
eg
or
y
ar
e
det
er
m
i
ned
b
y
es
t
abl
i
s
hi
ng
f
u
z
z
y
s
i
m
i
l
ar
m
at
r
i
x
.
F
u
zz
y
c
-
m
eans
al
g
or
i
t
hm
c
an m
a
k
e t
he s
am
pl
e
po
i
nt
s
of
v
ec
t
or
s
pac
e ac
c
or
di
ng t
o
a c
er
t
ai
n di
s
t
anc
e
m
eas
ur
em
ent
,
but
i
t
c
annot
f
ul
l
y
r
es
ear
c
h t
he
pi
x
el
s
’
gr
a
y
s
c
a
l
e c
h
ar
ac
t
er
i
s
t
i
c
s
an
d t
h
e c
or
r
el
at
i
on
be
t
w
een
adj
ac
e
nt
p
i
x
e
l
s
,
i
t
m
ak
es
t
he f
u
z
z
y
c
-
m
eans
al
gor
i
t
hm
f
ai
r
l
y
s
ens
i
t
i
v
e t
o
no
i
s
e.
I
n a
ddi
t
i
on,
t
he f
u
z
z
y
c
-
m
eans
al
g
or
i
t
hm
i
s
f
ai
r
l
y
s
ens
i
t
i
v
e t
o i
n
i
t
i
al
c
l
us
t
er
i
n
g c
ent
er
and
t
h
e i
nput
c
l
a
s
s
i
f
i
c
at
i
on n
um
ber
,
s
o ada
pt
i
v
e m
edi
a
n
f
i
l
t
er
i
n
g i
s
pr
opos
e
d t
o e
l
i
m
i
nat
e
t
he
no
i
s
e,
t
he f
u
z
z
y
s
i
m
i
l
ar
r
el
at
i
on
al
gor
i
t
hm
c
l
us
t
er
i
ng
r
es
ul
t
i
s
us
ed t
o d
et
er
m
i
ne t
he
i
n
i
t
i
al
c
l
us
t
er
i
ng c
e
nt
er
and
t
h
e num
ber
of
c
at
e
go
r
i
es
of
t
he
f
u
z
z
y
c
-
m
eans
al
gor
i
t
hm
.
F
u
zz
y
c
-
m
eans
m
ak
e t
he dat
a s
et
12
{
,
,
.....,
}
=
n
X
xx
x
,
i
t
c
ons
i
s
t
s
b
y
t
h
e s
am
pl
es
,
and
t
he d
i
m
ens
i
on i
s
“
p”
t
o
“c
”
c
l
as
s
es
.
T
he i
ni
t
i
al
c
l
u
s
t
er
i
ng c
ent
er
s
m
at
r
i
x
“
U”
i
s
m
at
r
i
x
t
hat
c
ons
i
s
t
ed b
y
t
he pi
x
e
l
s
m
e
di
a
n
of
eac
h c
at
egor
y
.
I
f
t
he
“u
ik
”
i
s
t
he
m
e
m
ber
s
hi
p d
egr
ee m
at
r
i
x
el
em
ent
b
et
w
ee
n
“x
i
”
and
“x
k
”
,
t
he r
es
ul
t
of
f
u
zz
y
c
-
m
eans
c
l
us
t
er
i
n
g i
s
r
epr
es
ent
e
d b
y
f
u
z
z
y
m
e
m
ber
s
hi
p degr
e
e m
at
r
i
x
.
T
he f
uz
z
y
m
e
m
ber
s
hi
p de
gr
ee m
at
r
i
x
i
s
s
ho
w
n
in
t
he equ
at
i
on be
l
o
w
:
()
()
2
1
()
1
1
()
−
=
=
∑
t
ij
t
c
ij
l
t
k
k
j
u
d
u
(
9)
“d
ij
”
i
s
t
he
di
f
f
er
enc
e b
et
w
een
t
he
c
l
us
t
er
i
n
g c
ent
er
of
num
ber
”
i”
an
d t
h
e s
a
m
pl
e of
num
ber
“
j
”
,
“
l”
i
s
t
he
w
e
i
gh
t
ed
i
nd
ex
.
T
he
c
l
us
t
er
i
ng
c
ent
er
n
eeds
t
o be
c
ons
t
ant
l
y
i
t
er
a
t
ed.
C
l
us
t
er
i
ng c
e
nt
er
s
i
t
er
at
i
on
f
or
m
ul
a i
n t
hi
s
p
aper
i
s
s
ho
w
n
i
n
t
h
e eq
uat
i
on
be
l
o
w
:
()
()
()
()
1:
1
1
(
)/
=
=
=
=
⋅⋅
⋅
∑∑
∑
mn
m
t
tt
t
ij
i
P
uU
f
I
u
(
10)
”f
”
i
s
t
he
m
at
r
i
x
of
or
i
gi
n
al
i
m
age ,
“U
(t)
”
i
s
t
he
i
m
age p
ar
t
i
t
i
on m
at
r
i
x
i
t
er
at
ed
”
t”
t
i
m
es
.
“
I”
is
t
h
e
m
a
t
r
ix
w
it
h
“m
”
l
i
nes
and
on
e c
o
l
um
n t
hat
eac
h e
l
em
ent
i
s
1.
T
he f
ol
l
o
w
i
ng f
i
gur
e
s
ar
e
c
l
us
t
er
i
n
g r
es
ul
t
s.
F
i
gur
e
4
.
T
he or
i
gi
nal
i
m
age
F
i
gur
e 5.
F
u
z
z
y c
-
m
eans
bas
ed I
m
pr
ov
e
d
m
a
k
er
w
at
er
ed
s
e
gm
ent
at
i
on
3.
R
e
su
l
t
s
a
n
d
A
n
a
l
y
s
i
s
I
S
O
D
A
T
A
a
l
gor
i
t
hm
i
s
one
of
t
he c
l
us
t
er
i
ng
al
g
or
i
t
hm
,
t
he c
l
us
t
er
i
n
g pr
oc
es
s
i
s
r
e
al
i
z
e
d
b
y
c
ons
t
ant
l
y
i
t
er
a
t
i
o
n.
S
a
m
pl
es
c
an be t
r
ans
f
er
r
ed f
r
o
m
a pol
y
m
er
i
z
at
i
on c
l
as
s
t
o anot
her
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
F
u
z
zy
C
-
M
ea
ns
C
l
us
t
er
i
n
g
B
as
ed
on
I
mpr
ov
ed
M
ar
k
e
d
W
at
er
s
hed…
(
C
u
ij
ie
Z
h
a
o
)
985
I
S
O
D
A
T
A
c
l
us
t
er
i
ng a
l
gor
i
t
hm
needs
t
o c
ons
t
ant
l
y
adj
us
t
t
he c
l
us
t
er
i
ng r
es
ul
t
s
t
hr
ough t
h
e
c
ont
i
n
uous
i
t
er
at
i
v
e c
l
us
t
er
i
ng c
en
t
er
.
I
S
O
D
A
T
A
al
gor
i
t
hm
i
s
v
er
y
s
e
ns
i
t
i
v
e
t
o t
he i
n
i
t
i
al
c
l
us
t
er
i
n
g
c
ent
er
,
but
t
h
er
e
i
s
m
uc
h
noi
s
e
i
nf
or
m
at
i
on,
and
m
an
y
par
am
et
er
s
nee
d
t
o
be
s
et
up
i
n t
hi
s
al
g
or
i
t
hm
.
I
m
age c
l
as
s
i
f
i
c
at
i
on
bas
e
d o
n I
S
O
D
A
T
A
al
gor
i
t
hm
i
s
s
ho
w
n
i
n t
he
f
ol
l
o
w
i
n
g
f
i
gur
e.
F
i
gur
e
6.
I
S
O
D
A
T
A
c
l
us
t
er
i
ng r
es
ul
t
T
he r
oad
ar
ea
i
n
t
h
e
i
m
a
ge t
hat
m
ar
k
ed A
i
s
di
v
i
d
ed
i
nt
o f
ar
m
i
n F
i
gur
e 6
,
but
i
t
i
s
di
v
i
d
ed
i
nt
o r
oa
d i
n F
i
gur
e
5.
A
not
h
er
ex
am
pl
e
i
s
t
he
pr
oc
es
s
i
ng
of
F
i
gur
e
7,
f
i
gu
r
e 8 a
nd f
i
gur
e
9
ar
e
t
he
r
es
ul
t
s
of
t
he
c
l
as
s
i
f
i
c
at
i
on
,
t
he
c
ont
r
as
t
i
s
v
er
y
ob
v
i
ous
.
t
hr
o
ugh
c
o
m
par
at
i
v
e
ex
per
i
m
ent
s
,
i
t
c
an
be s
e
e
n t
he
i
m
pr
ov
ed
al
gor
i
t
h
m
i
s
m
or
e ac
c
ur
at
e t
h
an
t
he
I
S
O
D
A
T
A
al
g
or
i
t
hm
.
O
pt
i
m
i
z
ed a
l
g
or
i
t
hm
not
onl
y
el
i
m
i
nat
es
t
he
i
m
pul
s
e noi
s
e
ef
f
ec
t
i
v
el
y
,
b
ut
al
s
o
av
oi
ds
ov
er
-
s
egm
ent
at
i
on
pr
ob
l
em
and r
e
duc
e t
he
no
i
s
e on
t
h
e c
l
us
t
er
i
ng.
F
i
gur
e
7
.
O
r
ig
in
a
l
i
m
age
F
i
gur
e
8
.
I
SO
D
AT
A c
l
u
s
t
e
r
i
n
g
F
i
gur
e
9
.
F
u
zzy
c
-
m
eans
c
lu
s
t
e
r
in
g
4
.
C
o
n
c
l
u
s
i
o
n
T
hi
s
paper
m
ai
nl
y
i
nt
r
od
uc
es
t
he f
u
z
z
y
C
-
m
eans
c
l
u
s
t
er
i
ng
opt
i
m
i
z
at
i
on
al
g
or
i
t
hm
.
I
t
s
t
ar
t
s
f
r
o
m
t
he
i
n
t
r
oduc
t
i
on
of
bas
i
c
t
he
or
i
es
,
and
gr
a
dua
l
l
y
ex
pl
or
es
t
he
p
ar
t
i
c
l
e
m
ar
k
er
-
bas
ed
w
at
er
s
he
d s
egm
ent
at
i
o
n.
A
nd t
h
en e
l
ab
or
at
es
t
he
w
hol
e al
g
or
i
t
hm
b
y
ana
l
y
z
i
n
g t
he pr
os
an
d
c
ons
of
t
he
al
g
or
i
t
hm
and
i
nt
r
oduc
i
ng
t
he
a
l
gor
i
t
hm
pr
oc
edur
e.
I
n
or
der
t
o
v
er
i
f
y
t
he
s
uper
i
or
i
t
y
of
t
he al
gor
i
t
hm
pr
opos
ed
her
e
，
t
hi
s
p
a
p
er
a
d
o
p
t
s
t
he f
u
z
z
y
c
-
m
eans
c
l
us
t
er
i
ng opt
i
m
i
z
at
i
on
al
g
or
i
t
hm
and
t
he
I
S
O
D
A
T
A
al
gor
i
t
hm
t
o
r
es
pec
t
i
v
e
l
y
cl
a
ssi
f
y
sa
m
p
l
e
s.
T
hr
ough
t
he
c
o
m
par
at
i
v
e
s
i
m
ul
at
i
on
ex
p
er
i
m
ent
s
,
i
t
c
an
be
s
e
en
f
r
om
t
he
ex
per
i
m
ent
dat
a
a
n
al
y
s
i
s
t
hat
t
he
f
u
z
z
y
c
-
m
eans
c
l
us
t
er
i
ng
o
pt
i
m
i
z
at
i
on
a
l
gor
i
t
hm
c
anno
t
on
l
y
i
m
pr
ov
e
c
la
s
s
if
ic
a
t
i
o
n
ac
c
ur
ac
y
,
b
ut
al
s
o ac
c
el
er
at
e
t
he c
o
nv
er
genc
e
s
pe
ed
and
enh
anc
e
t
he
a
bi
l
i
t
y
of
a
v
oi
di
ng
l
oc
a
l
ex
t
r
em
es
.
T
he
r
es
ul
t
s
s
ho
w
t
hat
t
he m
et
hod pr
opos
e
d i
n t
h
i
s
pa
p
er
i
s
m
or
e ac
c
ur
at
e t
han
t
he I
S
O
D
A
T
A
m
et
hod,
and
i
t
i
s
a f
eas
i
bl
e
t
r
ai
n
i
ng m
et
hod
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1
6
9
3
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
14
,
N
o
.
3,
S
ept
em
ber
2016
:
9
81
–
9
86
986
A
c
k
n
o
w
l
e
d
g
e
m
e
n
ts
T
hi
s
w
or
k
w
as
s
uppor
t
e
d b
y
t
h
e r
es
ear
c
h pr
oj
e
c
t
s
of
t
he s
oc
i
al
s
c
i
enc
e and
hum
ani
t
y
on
Y
o
ung F
und
of
t
he m
i
ni
s
t
r
y
of
E
d
uc
at
i
on
15
Y
J
C
Z
H
208
.
R
ef
er
en
ces
[1
]
Li
qi
an C
h
eng.
R
e
m
ot
e
s
ens
i
ng i
m
ag
e
c
l
a
s
s
i
f
i
c
at
i
on
ba
s
e
d on
op
t
i
m
i
z
ed
s
up
por
t
v
ec
t
or
m
ac
h
i
ne
.
T
E
LK
O
M
N
I
K
A
I
ndone
s
i
a
n J
ou
r
nal
o
f
E
l
e
c
t
r
i
c
al
E
ngi
neer
i
ng
.
2014;
12(
2)
:
10
37
-
10
45.
[2
]
B
S
ai
C
ha
nda
na.
C
l
us
t
er
i
n
g
A
l
gor
i
t
hm
C
om
bi
n
ed w
i
t
h H
i
l
l
C
l
i
m
bi
ng
f
or
C
l
as
s
i
f
i
c
a
t
i
o
n
of
R
e
m
ot
e
S
ens
i
ng I
m
ag
e
.
I
n
t
er
na
t
i
o
nal
J
our
nal
of
E
l
ec
t
r
i
c
al
a
nd C
om
p
ut
er
E
ng
i
ne
er
i
n
g
.
20
14;
4(
6)
:
9
23
-
930
.
[3
]
Y
Z
ar
negar
ni
a
,
H
A
Ma
j
d
,
M
R
T
av
i
r
ani
,
N
K
hai
er
,
A
A
K
M
a
boodi
.
A
ppl
i
c
a
t
i
on
of
f
uz
z
y
c
l
us
t
er
i
ng i
n
anal
y
s
i
s
of
i
nc
l
ud
ed
pr
o
t
ei
ns
i
n
es
o
phag
us
,
s
t
o
m
a
c
h
an
d
c
ol
on
c
an
c
er
s
ba
s
ed
o
n
s
i
m
i
l
ar
i
t
y
of
G
ene
O
nt
ol
ogy
a
nnot
at
i
o
n
.
K
oom
e
s
h
.
201
0;
1
2(
01)
:
14
-
22
.
[4
]
B
el
i
a
k
ov
G
,
J
am
e
s
S
,
Li
G
.
Lear
i
ng
c
ho
quet
-
i
nt
e
gr
al
-
bas
ed
m
at
r
i
c
es
f
or
s
em
i
s
u
per
v
i
s
ed
c
l
us
t
er
i
n
g
.
I
E
E
E
T
r
an
s
ac
t
i
on
s
o
n F
uz
z
y
S
y
s
t
em
s
.
20
11;
19(
3)
:
562
-
574.
[5
]
J
ul
i
an
F
an,
J
i
ng
Li
.
A
F
i
x
ed S
uppr
e
s
s
e
d R
at
e S
e
l
ec
t
i
on
M
et
hod f
or
S
u
ppr
e
s
s
e
d F
uz
z
y
C
-
M
eans
C
l
us
t
er
i
ng A
l
gor
i
t
h
m
.
A
p
pl
i
ed M
at
hem
at
i
c
s
.
2
014;
5(
8)
:
127
5
-
1283.
[6
]
Ji
Z
e
-
x
uan
,
C
he
n
Q
i
an
g
,
S
un
Q
uan
-
s
en
,
Xi
a
De
-
s
hen
.
I
m
age
S
eg
m
ent
at
i
o
n
wi
t
h
A
n
is
o
t
r
opi
c
W
eig
ht
e
d
F
u
zzy
C
-
M
eans
C
l
us
t
er
i
ng
.
J
o
ur
nal
o
f
C
om
put
er
A
i
d
ed D
e
s
i
gn &
C
om
put
er
G
r
aph
i
c
s
.
20
09;
21(
1
0)
:
1451
-
1
459.
[7
]
W
a
n
g X
i
ao
peng
,
H
ao
C
ho
ng
y
ang,
F
a
n Y
angy
u.
W
a
t
er
s
he
d S
eg
m
ent
at
i
on B
as
e
d o
n M
or
phol
ogi
c
a
l
Sc
a
l
e
-
S
pac
e
an
d
G
r
adi
e
nt
M
o
di
f
i
c
at
i
on
.
J
our
n
al
of
E
l
ec
t
r
on
i
c
s
&
I
nf
or
m
at
i
on
T
ec
h
nol
o
gy
.
2
006
;
2
8
(3
):
485
-
48
9.
[8
]
G
ao
Li
,
Y
ang
S
huy
an,
X
i
a
J
i
e,
e
t
al
.
A
N
ew
M
ar
k
er
-
B
a
s
ed W
a
t
er
s
hed
A
l
gor
i
t
hm
.
A
c
ta
E
l
ec
t
r
oni
c
a
S
in
ic
a
.
2
006
;
34(
1
1)
:
201
8
-
20
23.
[9
]
B
as
r
an P
ar
m
i
nder
S
,
R
ober
t
s
on A
ndr
ew
,
W
e
l
l
s
D
er
e
k
.
C
T
i
m
ag
e ar
t
i
f
ac
t
s
f
r
om
br
ac
hy
t
h
er
apy
s
ee
d
i
m
pl
ant
s
:
a
po
s
t
pr
o
c
e
s
s
i
ng
3
D
adapt
i
v
e m
e
di
a
n f
i
l
t
er
.
M
edi
c
al
P
hy
s
i
c
s
.
2
011
;
38(
2)
:
7
12
-
718.
Evaluation Warning : The document was created with Spire.PDF for Python.