T
E
L
KO
M
NI
K
A
,
V
ol
.
14,
N
o.
3,
S
ept
em
ber
20
16,
pp.
11
75
~
118
2
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
.
3292
11
75
R
ec
ei
v
ed
D
ec
e
mber
18
,
20
1
5
;
R
e
v
is
ed
J
u
ne 2
1
,
2
01
6
;
A
c
c
ept
e
d
J
un
e
10
,
20
1
6
Im
ag
e Ret
r
ie
v
a
l Based
o
n
M
ul
ti
S
tr
uc
t
ur
e
Co
-
o
ccu
r
r
e
n
ce
Descr
ip
t
o
r
Ag
u
s
Ek
o
M
i
n
a
r
n
o
*
1
,
A
r
r
i
e
K
u
r
n
i
a
w
a
r
d
h
a
n
i
2
,
F
i
tr
i
B
i
m
a
n
to
r
o
3
1
U
ni
v
er
s
i
t
as
M
uham
m
ad
i
y
ah
M
al
ang,
M
al
ang
,
I
nd
one
s
i
a
2
U
ni
v
er
s
i
t
as
I
s
l
am
I
ndo
nes
i
a,
Y
ogy
ak
ar
t
a
,
I
nd
o
n
es
ia
3
U
ni
v
er
s
i
t
as
M
at
ar
am
,
M
at
ar
am
,
I
nd
o
n
es
ia
*
C
or
r
es
po
ndi
ng
a
ut
hor
,
e
-
m
a
i
l:
agoe
s
.
m
i
nar
no@
gm
a
il
.
c
o
m
A
b
st
r
act
T
hi
s
s
t
ud
y
pr
es
e
nt
s
a
new
t
e
c
hni
que
f
or
B
at
i
k
c
l
ot
h
i
m
age
r
et
r
i
ev
a
l
u
s
i
ng
M
i
c
r
o
-
St
ru
c
t
u
r
e
C
o
-
oc
c
ur
en
c
e
D
es
c
r
i
pt
or
(
M
S
C
D
)
.
M
S
C
D
i
s
one
of
f
eat
ur
e
e
x
t
r
a
c
t
i
o
n
m
et
hod.
I
t
i
s
a
dev
e
l
op
e
d
m
et
hod
bas
ed
on E
nhan
c
e
d
M
i
c
ro
St
ru
c
t
u
re
D
e
s
c
ri
p
t
o
r (EM
SD
)
.
I
n pr
ev
i
ou
s
s
t
ud
y
,
E
M
S
D
has
been us
e
d f
or
ex
t
r
ac
t
i
n
g
t
ex
t
ur
e,
c
o
l
or
,
and
s
hap
e
f
eat
ur
e
s
i
m
ul
t
aneo
us
l
y
.
H
ow
ev
er
,
E
M
S
D
c
oul
d
not
ac
h
i
ev
e
an
o
pt
i
m
al
pr
e
c
i
s
i
o
n
f
or
B
at
i
k
c
l
ot
h i
m
age r
et
r
i
ev
a
l
.
T
her
ef
or
e,
M
S
C
D
i
s
pr
opos
ed t
o
o
v
er
c
om
e t
he E
M
S
D
dr
aw
bac
k
us
i
n
g
gl
oba
l
f
eat
ur
e ap
pr
oa
c
h,
nam
el
y
G
r
a
y
L
ev
e
l
C
o
-
o
c
c
ur
r
en
c
e M
at
r
i
x
(
G
LC
M
)
.
T
her
e ar
e
300 b
at
i
k
c
l
ot
h
i
m
ages
,
w
hi
c
h c
ont
a
i
n
50 c
l
as
s
es
us
e
d f
or
dat
a
s
et
.
T
he
per
f
or
m
an
c
e r
es
u
l
t
s
how
s
t
hat
M
S
C
D
c
an
i
n
c
r
ea
s
e
p
r
ec
is
i
on
a
nd
r
ec
a
ll
,
by
1
9.
03
%
a
nd
1
8.
33
%,
r
e
s
pec
t
i
v
e
l
y
.
I
t
i
ndi
c
at
es
t
hat
M
S
C
D
c
an
r
e
t
r
i
e
v
e
B
at
i
k
c
l
ot
h i
m
a
ges
m
or
e ef
f
ec
t
i
v
e t
h
an E
M
S
D
.
Ke
y
w
o
rd
s
:
b
at
i
k
,
i
ma
g
e r
e
t
r
iev
a
l
,
m
i
c
ro
s
t
ru
c
t
u
re
,
g
r
ay
le
v
el
c
o
-
o
c
c
ur
r
enc
e m
at
r
ix
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
B
at
i
k
i
s
a w
or
d c
am
e f
r
o
m
J
av
anes
e t
hat
c
ons
i
s
t
of
“
ngem
bat
”
and “
ni
t
i
k
”
[
1
]
.
F
ro
m
t
hos
e
s
t
em
s
,
B
at
i
k
i
s
a
t
r
a
di
t
i
on
al
m
et
hod
t
o
dr
a
w
a
m
ot
i
f
on
c
l
ot
h
[
2
-
3]
.
B
at
i
k
i
s
ac
k
now
l
e
dge
d
b
y
U
N
E
S
C
O
as
one of
I
ndones
i
an c
ul
t
ur
al
h
er
i
t
a
ge
o
n 2 O
c
t
ober
200
9.
P
at
t
er
ns
c
o
m
pos
ed
B
a
t
ik
c
lo
t
h
is
a
s
y
m
bo
l
t
h
at
has
a par
t
i
c
ul
ar
m
eani
n
g,
es
pec
i
al
l
y
f
or
J
av
a
nes
e
peo
pl
e
[
3
-
4]
.
T
her
e ar
e s
o
m
e s
pec
i
f
i
c
pat
t
er
ns
t
h
at
m
us
t
be us
ed i
n a s
pec
i
f
i
c
m
o
m
ent
and
b
y
a s
pec
i
f
i
c
per
s
on.
B
at
i
k
pat
t
er
ns
ar
e
c
o
m
pos
ed
r
ep
eat
edl
y
and
ei
t
h
er
g
eom
et
r
i
c
al
l
y
or
no
n
-
geom
et
r
i
c
al
l
y
.
A
c
c
or
di
n
g t
o t
hos
e
pat
t
er
ns
,
B
at
i
k
i
m
ages
ar
e r
ec
ogn
i
z
ed
i
n or
d
er
t
o f
ac
i
l
i
t
at
e o
n m
a
k
i
ng
doc
um
ent
at
i
o
n.
S
om
e s
t
udi
es
ha
v
e
be
en p
r
opos
ed t
o r
ec
ogn
i
z
e
B
at
i
k
i
m
ages
.
I
da
,
et
al
.,
[
1]
h
av
e
done
per
f
or
m
anc
e c
o
m
par
i
s
on on t
hr
ee f
eat
ur
e ex
t
r
ac
t
i
on m
et
hods
,
nam
el
y
G
r
a
y
L
ev
el
C
o
-
oc
c
ur
r
enc
e Mat
r
i
c
es
(
G
LC
M)
,
C
an
n
y
E
dge
D
et
ec
t
i
o
n,
and
G
abor
f
i
l
t
er
s
.
T
hat
s
t
u
d
y
s
h
o
w
e
d t
h
at
G
LC
M i
s
s
up
er
i
or
t
o C
a
nn
y
E
dge
D
et
ec
t
i
on
and
G
a
bor
f
i
l
t
er
s
w
he
n i
t
w
as
us
e
d t
o r
epr
es
e
nt
B
at
i
k
i
m
ages
.
C
heong an
d
Lok
e [
5]
us
ed
T
c
hebi
c
hef
O
r
t
hogona
l
P
o
l
y
nom
i
al
s
t
o obt
a
i
n s
hap
e
f
eat
ur
e of
S
o
ngk
et
and
B
at
i
k
i
m
age f
r
o
m
Mul
t
i
s
pec
t
r
al
Co
-
oc
c
ur
r
enc
e M
a
t
r
i
c
e
s
.
T
hat
s
t
ud
y
i
nd
i
c
at
e
d t
h
at
T
c
hebi
c
hef
O
r
t
hogo
na
l
P
ol
y
nom
i
al
s
c
an
be us
ed
t
o
i
m
pr
ov
e t
he
ac
c
ur
ac
y
of
B
a
t
i
k
C
ol
or
i
m
age r
ec
ogni
t
i
o
n.
Lok
e and C
he
ong
[
6
]
h
ad
done
f
eat
ur
e r
educ
t
i
on
u
s
i
ng
P
r
i
nc
i
pal
C
omp
one
nt
A
na
l
y
s
i
s
(
P
C
A
)
.
T
hat
s
t
ud
y
s
ho
w
e
d
t
hat
P
C
A
c
an
r
e
duc
e f
eat
ur
es
pr
ec
i
s
el
y
,
s
o
i
t
c
an hel
p t
o
m
a
k
e
r
ec
ogni
t
i
on
f
as
t
er
w
i
t
ho
ut
r
edu
c
i
ng t
he ac
c
ur
ac
y
.
R
a
ng
k
ut
i
,
et
al
.,
[7
]
dev
el
ope
d
C
o
nt
e
nt
-
bas
e
d
I
mage
R
et
r
i
ev
a
l
(
C
B
I
R
)
f
or
B
at
i
k
i
m
age
us
i
ng
c
a
nn
y
e
dge
det
ec
t
i
o
n
and
s
hap
e
i
n
v
ar
i
ant
m
o
m
ent
f
or
ex
t
r
ac
t
i
ng
t
ex
t
ur
e
and
s
hap
e
f
eat
ur
e,
r
es
pe
c
t
i
v
e
l
y
.
T
hos
e
s
t
udi
es
,
h
o
w
e
v
er
,
pr
o
v
i
de
d uns
at
i
s
f
i
ed r
es
ul
t
f
or
B
at
i
k
i
m
ages
t
hat
ha
v
e m
o
r
e c
o
m
pl
i
c
at
e
d
pat
t
er
n
w
h
i
c
h c
on
t
ai
ns
v
ar
i
ous
pa
t
t
er
ns
i
n
a B
at
i
k
i
m
age.
A
s
a c
o
ns
equ
enc
e,
B
a
t
i
k
i
m
ages
i
nt
er
-
c
l
as
s
hav
e s
i
m
i
l
ar
pat
t
er
n
[
1]
.
B
ec
aus
e of
t
hat
,
a m
or
e r
el
i
a
bl
e f
eat
ur
e ex
t
r
ac
t
i
on m
et
hod i
s
need
ed.
T
her
e ar
e s
om
e
f
eat
ur
e e
x
t
r
ac
t
i
on m
et
hods
t
ha
t
c
om
bi
ne t
ex
t
ur
e,
c
ol
or
,
an
d
s
hape
f
eat
ur
e
b
as
ed
on
c
o
-
oc
c
ur
r
enc
e
m
at
r
i
c
es
,
nam
el
y
T
ex
t
on
C
o
-
oc
c
ur
r
enc
e
Ma
t
r
i
x
[
9]
,
Mu
l
t
i
-
T
ex
t
on
H
i
s
t
ogr
am
[
10]
,
Mi
c
r
o
-
S
t
r
uc
t
ur
e
D
es
c
r
i
p
t
or
(
M
S
D
)
[
11]
,
C
o
l
or
D
i
f
f
er
enc
e
H
i
s
t
ogr
am
and
G
LC
M [
1
2]
,
M
ul
t
i
T
ex
t
on C
o
-
oc
c
ur
r
enc
e D
es
c
r
i
pt
or
(
MT
C
D
)
[
13]
,
W
av
el
et
an
d G
LC
M [
14]
,
E
nh
anc
ed
Mi
c
r
o
-
S
t
r
uc
t
ur
e
D
es
c
r
i
pt
or
(
E
M
S
D
)
[
15
]
,
G
eom
et
r
i
c
F
eat
ur
e
E
x
t
r
ac
t
i
on
[
1
6]
,
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
:
11
75
–
1
182
1176
C
l
as
s
i
f
i
c
at
i
on us
i
n
g S
I
F
T
[
18]
,
et
c
.
E
M
S
D
i
s
on
e of
r
el
i
a
bl
e,
s
i
m
pl
e,
and ef
f
i
c
i
ent
m
et
hod f
or
ex
t
r
ac
t
i
ng f
eat
ur
es
.
E
M
S
D
ex
t
r
ac
t
t
ex
t
ur
e,
c
ol
or
,
and
s
hape f
eat
ur
e s
i
m
ul
l
t
an
e
o
us
l
y
b
as
ed o
n
m
i
cr
o
-
s
t
uc
t
ur
e
and
e
dge
o
r
i
ent
a
t
i
o
n s
i
m
i
l
ar
i
t
y
b
ut
E
M
S
D
c
a
nnot
ac
h
i
e
v
e
an
op
t
i
m
al
pr
ec
i
s
i
on.
T
her
ef
or
e,
Mi
c
r
o
-
S
t
r
uc
t
ur
e
C
o
-
oc
c
ur
r
enc
e D
es
c
r
i
pt
or
(
MS
C
D
)
i
s
pr
opos
e
d t
o
ov
er
c
om
e
t
he
E
M
S
D
dr
a
w
bac
k
us
i
ng
gl
o
bal
f
eat
ur
e a
ppr
oac
h,
n
am
e
l
y
G
LC
M.
2.
D
at
aset
B
at
i
k
i
m
age dat
as
et
i
s
c
ol
l
ec
t
ed b
y
c
apt
ur
i
ng 5
0 t
y
p
es
of
B
at
i
k
c
l
ot
h.
E
ac
h c
l
ot
h i
s
c
apt
ur
ed
t
o
as
m
uc
h
as
s
i
x
r
andom
i
m
ages
and
t
h
en
r
es
i
z
ed t
o 1
28x
12
8
p
i
x
el
s
s
i
z
e
i
n
J
P
E
G
f
or
m
at
.
T
hus
,
t
he t
ot
al
nu
m
ber
of
i
m
ages
i
n a dat
a
s
et
i
s
300 and c
ons
i
s
t
s
o
f
50 c
l
as
s
es
.
E
x
am
pl
es
of
bat
i
k
i
m
ages
ar
e
s
ho
w
n
i
n
F
i
g
ur
e
1.
I
n
g
ener
a
l
,
t
her
e
ar
e
t
w
o
p
at
t
er
ns
of
c
apt
ur
ed
bat
i
k
i
m
ages
,
geom
et
r
i
c
an
d no
n
-
geom
et
r
i
c
pat
t
er
ns
.
F
i
gur
e
1.
E
x
am
pl
e of
B
at
i
k
i
m
ages
3.
M
u
l
ti
S
tr
u
c
tu
r
e
C
o
-
o
c
cu
r
r
en
ce D
esc
r
i
p
t
o
r
MS
C
D
i
s
a
de
v
e
l
op
ed
m
et
ho
d
b
as
ed
on
E
MS
D
t
hat
i
s
c
om
bi
ned
w
i
t
h
G
L
C
M
t
o
ov
er
c
om
e E
M
S
D
dr
a
w
bac
k
.
E
MS
D
h
as
78
f
eat
ur
es
w
hi
c
h c
o
ns
i
s
t
of
c
ol
or
f
eat
ur
e an
d e
dge
or
i
ent
a
t
i
on
d
er
i
v
ed
f
r
om
m
i
c
r
o
-
s
t
r
uc
t
ur
e m
ap
of
edge
det
ec
t
i
on.
W
her
eas
G
LC
M
has
16
f
eat
ur
e
as
gl
o
ba
l
f
eat
ur
e r
e
pr
es
ent
at
i
o
n of
an
i
m
age.
F
i
gur
e
2.
F
e
at
ur
es
ex
t
r
ac
t
i
o
n di
agr
am
of
M
ic
r
o
-
S
t
r
uc
t
ur
e
Co
-
O
c
c
ur
r
enc
e D
es
c
r
i
pt
o
r
A
c
c
or
di
n
g t
o
F
i
g
ur
e 2
,
f
i
r
s
t
s
t
age
of
MS
C
D
i
s
c
on
v
e
r
t
i
ng
R
G
B
i
m
age t
o
H
S
V
i
m
age.
T
hen,
t
hat
H
S
V
i
m
age
i
s
q
uant
i
z
ed
i
nt
o
8
bi
ns
f
or
H
,
3
bi
ns
f
or
S
,
a
nd
3
b
i
ns
f
or
V
.
T
hus
,
t
h
er
e
ar
e 8x
3x
3
=
72
c
ol
or
f
eat
ur
e
s
.
S
ec
on
d
s
t
age
of
MS
C
D
i
s
edge
or
i
e
nt
at
i
o
n
det
ec
t
i
on
us
i
n
g
S
o
bel
o
per
at
or
.
E
dge
or
i
net
at
i
on s
t
r
et
c
hs
f
r
o
m
0º
t
o 18
0º
.
T
hen,
t
h
at
edg
e
or
i
ent
at
i
on
i
s
qua
nt
i
z
e
d i
n
t
o 6 or
i
e
nt
a
t
i
o
n
bi
ns
w
i
t
h
30
º
as
t
h
e s
t
ep
-
l
e
ngt
h f
or
eac
h
b
i
n.
T
hus
,
t
h
e
r
e ar
e 6
edg
e or
i
ent
at
i
on f
e
at
ur
es
.
T
hi
r
d s
t
age
i
s
bui
l
d
i
n
g a
m
i
c
r
o
-
s
t
r
uc
t
ur
e m
ap f
r
o
m
edge or
i
ent
at
i
on
i
m
age t
h
at
ha
s
bee
n
quan
t
i
z
ed
t
o m
a
k
e a m
as
k
t
hat
w
i
l
l
b
e s
uper
i
m
pos
ed ov
er
H
S
V
i
m
age t
hat
has
be
en qu
ant
i
z
ed
.
A
m
ic
r
o
-
s
t
r
uc
t
ur
e
m
ap
i
s
bui
l
t
b
as
ed
o
n
or
i
ent
at
i
o
n
s
i
m
i
l
ar
i
t
y
det
ec
t
i
o
n
i
n
q
uant
i
z
ed
e
dg
e
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
I
mag
e R
et
r
i
ev
a
l
B
as
ed
on
Mul
t
i
S
t
r
uc
t
ur
e
Co
-
o
c
c
ur
r
en
c
e
D
es
c
r
i
pt
or
(
A
gus
Ek
o
Mi
nar
no
)
1177
or
i
ent
at
i
on i
m
age us
i
n
g 3x
3 gr
i
d k
er
nel
.
T
he or
i
ent
at
i
on v
a
l
u
es
t
hat
ar
e c
om
par
ed ar
e c
ent
er
gr
i
d o
r
i
e
nt
at
i
o
n of
k
er
nel
t
o 8
-
nei
ghb
or
gr
i
ds
of
k
er
nel
t
hat
i
s
1 gr
i
d di
s
t
anc
e f
r
o
m
t
he c
ent
er
.
T
he
m
i
cr
o
-
s
t
r
uc
t
ur
e
m
ap
i
s
ar
r
anged
f
r
om
pat
t
er
ns
w
her
e
t
he
or
i
ent
a
t
i
o
n
of
nei
ghb
or
gr
i
ds
ha
v
e
t
he
s
a
m
e v
a
l
ue
w
i
t
h t
he
or
i
en
t
at
i
o
n of
c
ent
er
gr
i
d.
T
hen
t
hat
pat
t
er
n
b
ec
om
e a m
a
s
k
t
hat
w
i
l
l
b
e
s
uper
i
m
pos
ed o
v
er
q
uant
i
z
ed H
S
V
i
m
age.
T
hi
s
s
t
age
i
s
i
l
l
us
t
r
at
e
d i
n F
i
g
ur
e 3(
a)
-
3(
g)
.
F
our
t
h s
t
age
i
s
bui
l
d
i
ng
a m
i
c
r
o
-
s
t
r
uc
t
ur
e
m
ap
f
r
o
m
H
S
V
i
m
age t
ha
t
has
be
en qu
ant
i
z
e
d
t
o m
a
k
e a m
as
k
t
hat
w
i
l
l
b
e
s
uper
i
m
pos
ed o
v
er
edg
e o
r
i
en
t
a
t
i
o
n i
m
age
t
hat
has
b
een qua
nt
i
z
e
d.
A
m
ic
r
o
-
s
t
r
uc
t
ur
e
m
ap i
s
b
ui
l
t
b
as
ed
on
i
nt
ens
i
t
y
s
i
m
i
l
ar
i
t
y
det
ec
t
i
o
n i
n
qua
nt
i
z
e
d H
S
V
i
m
age
us
i
ng
3x
3
gr
i
d
k
er
nel
.
T
he
i
nt
ens
i
t
y
v
al
u
es
t
h
at
ar
e
c
o
m
par
ed
ar
e
c
ent
er
gr
i
d
i
nt
e
ns
i
t
y
of
k
er
nel
t
o
8
-
ne
i
gh
bor
gr
i
ds
of
k
er
nel
t
h
at
i
s
1
gr
i
d
d
i
s
t
anc
e
f
r
om
t
he
c
ent
er
.
T
he
m
i
c
r
o
-
s
t
r
uc
t
ur
e
m
ap
i
s
ar
r
ange
d f
r
o
m
pat
t
er
ns
w
her
e t
he
i
nt
ens
i
t
y
of
nei
gh
bor
gr
i
ds
ha
v
e t
he s
am
e v
a
l
ue
w
i
t
h t
he
i
nt
e
ns
i
t
y
of
c
ent
er
gr
i
d.
T
hen t
h
at
pa
t
t
er
n b
ec
om
e
a m
as
k
t
hat
w
i
l
l
be s
up
er
i
m
pos
ed ov
er
quan
t
i
z
ed
ed
ge or
i
en
t
at
i
on
i
m
age.
T
hi
s
s
t
age i
s
i
l
l
us
t
r
a
t
ed
i
n F
i
gur
e
3
(
h)
-
3
(j
).
F
i
f
t
h
s
t
age
of
MS
C
D
i
s
ac
c
or
di
n
g
t
o
G
LC
M
i
s
app
l
i
ed
t
o
t
h
e
or
i
gi
na
l
i
m
age
[
1
7
]
.
G
LC
M
f
eat
ur
es
t
hat
ar
e
us
ed
ar
e
ener
g
y
,
ent
r
o
p
y
,
c
ont
r
as
t
,
a
nd
c
or
r
el
at
i
on
f
eat
ur
e
at
0
º
,
45
º
,
9
0º
,
a
n
d
135º
.
T
hus
,
t
h
er
e
ar
e 4x
4
=
1
6 G
L
C
M f
eat
ur
es
.
H
en
c
e,
T
he t
ot
al
f
at
ur
es
t
ha
t
MS
C
D
has
ar
e
72+
6+
16
=
9
4 f
eat
ur
e
.
3.
1.
H
S
V
C
o
l
o
r
S
p
a
c
e
an
d
C
o
l
o
r
Q
u
a
n
ti
z
a
ti
o
n
C
ol
or
pr
o
v
i
des
s
pat
i
a
l
c
or
r
el
at
i
on
i
nf
or
m
at
i
on t
hat
v
er
y
r
el
i
ab
l
e f
or
C
B
I
R
a
nd
obj
ec
t
r
ec
ogni
t
i
o
n
.
C
o
l
or
hi
s
t
o
gr
a
m
i
s
us
ed
f
r
equent
l
y
as
c
ol
or
f
eat
ur
e ex
t
r
ac
t
i
on m
et
ho
d i
n C
B
I
R
.
R
G
B
c
ol
or
s
pac
e i
s
eas
y
an
d pr
a
c
t
i
c
al
t
o b
e us
ed b
ut
c
ann
ot
r
epr
es
ent
hum
an v
i
s
ual
pe
r
c
ept
i
on
w
el
l
.
H
S
V
c
ol
or
s
p
ac
e i
s
bet
t
er
t
o r
epr
es
en
t
hum
an v
i
s
ua
l
p
er
c
ept
i
o
n t
h
an
R
G
B
.
HS
V
c
ol
or
s
pac
e c
o
ns
i
s
t
s
o
f
t
hr
ee c
om
ponent
s
,
nam
el
y
h
ue,
s
at
ur
at
i
o
n,
a
nd
v
a
l
u
e.
H
u
e
r
epr
es
ent
s
c
ol
or
b
as
ed
o
n
t
hei
r
d
om
i
nant
w
a
v
e
l
en
gt
h.
S
at
ur
a
t
i
o
n
r
epr
es
e
nt
s
t
he
pur
i
t
y
of
c
ol
or
or
t
he
am
ount
of
w
hi
t
e l
i
gh
t
s
m
i
x
ed w
i
t
h
a h
ue.
V
a
l
u
e r
epr
es
ent
s
t
he
br
i
gh
t
nes
s
of
c
ol
or
.
H
ue
(
H
)
r
anges
f
r
o
m
0º
t
o
360º
m
odel
e
d
i
n
a
c
i
r
c
l
e
f
or
m
,
w
hi
l
e
r
ed
at
0º
,
gr
een
at
12
0
º
and
b
l
ue
at
24
0
º.
S
at
ur
at
i
on
(
S
)
r
anges
f
r
o
m
0
t
o
1.
V
al
u
e
(
V
)
r
anges
f
r
om
0
t
o
1.
I
n
t
h
i
s
s
t
ud
y
,
w
e qu
ant
i
z
e H
S
V
c
ol
or
s
pa
c
e i
nt
o 7
2 b
i
ns
t
h
at
c
o
ns
i
s
t
of
8 b
i
ns
f
or
H
,
3 b
i
ns
f
or
S
,
and
3 b
i
ns
f
or
V
.
T
hus
,
H
S
V
i
m
age t
hat
h
as
bee
n qu
ant
i
z
ed
has
7
2 f
eat
ur
es
.
3.
2.
E
d
g
e
O
r
i
e
n
ta
ti
o
n
D
e
t
e
c
ti
o
n
i
n
H
SV
C
o
l
o
r
Sp
a
c
e
T
her
e
ar
e
v
ar
i
o
us
edg
e
det
ec
t
i
on
m
et
hods
,
s
uc
h
as
S
obe
l
,
R
ob
er
t
,
P
r
e
w
i
t
,
C
ann
y
,
an
d
LoG
o
per
at
or
.
S
obe
l
o
per
at
or
i
s
us
e
d i
n t
hi
s
s
t
u
d
y
t
hat
app
l
i
ed
at
e
ac
h c
om
ponen
t
of
H
S
V
c
ol
or
s
pac
e s
epar
at
el
y
.
I
n C
ar
t
e
s
i
an s
pac
e
t
he d
ot
pr
od
uc
t
of
v
ec
t
or
s
a(
x
1
,y
1
,z
1
)
an
d
b(
x
2
,y
2
,z
2
)
is
def
i
ned
as
:
ab
=
x
1
x
2
+y
1
y
2
+z
1
z
2
(
1)
c
os
(
,
)
=
|
|
|
|
=
1
2
+
1
2
+
1
2
1
2
+
1
2
+
1
2
2
2
+
2
2
+
2
2
(
2)
H
S
V
c
o
l
or
s
pac
e bas
e
d o
n c
y
l
i
nd
er
c
oor
di
n
at
e m
us
t
be t
r
ans
f
or
m
ed i
nt
o C
ar
t
es
i
an
c
oor
di
n
at
e.
L
et
(
H
,S
,V
)
ar
e
a poi
n
t
of
c
y
l
i
n
der
c
oor
d
i
n
at
e,
an
d (
H
’,
S
’,
V
’
)
ar
e
a po
i
nt
of
C
ar
t
es
i
an
c
oor
di
n
at
e,
s
o
t
h
at
H’
=
S
.
co
s(
H
)
,
S’
=S
.
si
n
(
H
)
,
and
V’
=
V
.
A
f
t
er
poi
nt
of
H
’,
S
’,
a
n
d
V
’
i
s
o
bt
a
i
ne
d,
S
ob
el
o
per
at
or
i
s
ap
pl
i
ed
i
n
eac
h
H
’,
S
’,
a
n
d
V
’
l
a
y
e
r
t
o
do
edge
d
et
ec
t
i
on.
E
dg
e
d
et
ec
t
i
on
y
i
e
l
ds
t
he
gr
a
di
e
nt
s
al
ong
x
an
d
y
di
r
ec
t
i
o
ns
t
hat
d
eno
t
ed
b
y
a(
H’
x
,
S
’
x
,
V
’
x
)
an
d
b(
H’
y
,
S
’
y
,
V
’
y
)
,
w
her
e
H’
x
i
s
gr
ad
i
ent
of
H’
al
o
ng h
or
i
z
ont
a
l
di
r
ec
t
i
on,
H’
y
i
s
gr
adi
ent
of
H’
a
l
on
g
v
er
t
i
c
al
d
i
r
ec
t
i
o
n,
a
nd s
o
on.
T
he do
t
pr
o
duc
t
of
v
ec
t
or
s
a(
H’
x
, S
’
x
, V
’
x
)
an
d b(
H’
y
, S
’
y
, V
’
y
)
,
i
s
def
i
ne
d as
:
|
|
=
(
’
)
2
+
(
’
)
2
+
(
’
)
2
(
3)
|
|
=
(
’
)
2
+
(
’
)
2
+
(
’
)
2
(
4)
=
′
′
+
′
′
+
′
′
(
5)
T
hus
,
t
he a a
ng
l
e
bet
w
een
v
ec
t
or
s
a(
H’
x
, S
’
x
, V
’
x
)
an
d
b(
H’
y
, S
’
y
, V
’
y
)
,
i
s
def
i
ne
d as
:
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
:
11
75
–
1
182
1178
c
os
(
,
)
=
|
|
|
|
(
6)
=
a
r
cco
s
(
)
=
a
r
cco
s
|
|
.
|
|
(
7)
Af
t
e
r
of
ed
ge
or
i
e
nt
at
i
on
i
s
obt
ai
ned,
i
t
i
s
q
uant
i
z
ed
i
nt
o
m
-
bi
ns
,
w
her
e
m
=
{
6,
12,
18,
2
4,
3
0,
3
6}
.
W
hi
l
e ed
ge
or
i
en
t
at
i
on
i
s
q
uan
t
i
z
ed
i
n
t
o s
i
x
bi
ns
,
eac
h
bi
n
has
30
º
as
t
he
s
t
ep
-
l
en
gt
h i
n
t
er
v
al
.
3.
3.
M
i
c
r
o
-
s
tr
u
c
tu
r
e
D
e
fi
n
i
ti
o
n
a
n
d
M
a
p
Ex
t
ra
c
ti
o
n
M
ic
r
o
-
s
t
r
uc
t
ur
e
of
an
i
m
age
i
s
f
oun
d
us
i
ng
3x
3
k
er
nel
c
onv
ol
v
e
d
on
t
o
t
hr
ou
gho
ut
pi
x
e
l
s
on i
m
age w
h
i
l
e l
ook
i
ng f
or
t
he s
i
m
i
l
ar
i
t
y
am
ong
t
he pi
x
el
t
hat
a
l
i
gne
d w
i
t
h
c
en
t
er
g
r
i
d of
k
er
nel
t
o
t
he
p
i
x
el
s
t
h
at
a
l
i
g
ned
w
i
t
h
8
-
nei
ghb
or
gr
i
ds
of
k
er
nel
t
hat
h
as
1
gr
i
d
di
s
t
a
nc
e
f
r
om
t
he
c
ent
er
.
P
ad
di
ng
i
s
add
ed
i
nt
o a
n i
m
age at
c
onv
ol
ut
i
on s
t
e
p.
K
er
ne
l
i
s
m
ov
ed
w
i
t
h
a s
t
e
p
-
l
en
gt
h
of
t
w
o
pi
x
el
s
f
r
om
r
i
ght
t
o l
ef
t
and f
r
om
t
op t
o bot
t
om
w
i
t
h di
f
f
er
ent
s
t
ar
t
i
n
g po
i
nt
.
S
t
ar
t
i
ng
poi
nt
s
us
ed i
n
t
hi
s
s
t
ud
y
ar
e
M1(
0,
0)
,
M2(
0,
1)
,
M3(
1,
0)
,
and
M4(
1,
1)
.
T
he
di
f
f
er
ent
s
t
ar
t
i
ng
p
oi
nt
i
s
us
ed t
o
av
o
i
d
an o
v
er
l
app
i
n
g.
T
he m
i
c
r
o
-
s
t
r
uc
t
ur
e
m
ap i
s
bui
l
t
i
n e
ac
h c
on
v
ol
ut
i
on f
r
om
pos
i
t
i
o
ns
of
pi
x
el
t
ha
t
has
t
he s
am
e
v
al
u
e
w
i
t
h t
h
e c
ent
er
p
i
x
el
.
T
he f
i
nal
m
i
c
r
o
-
s
t
r
uc
t
ur
e i
s
obt
ai
ne
d b
y
c
o
m
bi
ni
ng f
our
m
i
c
r
o
-
s
t
r
uc
t
ur
e m
ap,
M1,
M2,
M3,
and M4
,
b
as
ed
on f
ol
l
o
w
i
ng
r
ul
e
:
M
(
x
,
y)
=
U
N
IO
N
(M
1
(x
,
y
),
M
2
(x
,
y
),
M
3
(x
,
y
,
M
4
(x
,
y
)))
(
8)
1
1
4
3
4
4
1
1
4
3
4
4
1
1
4
3
4
4
1
1
4
3
4
s
2
2
4
3
1
2
2
2
4
3
1
2
2
2
4
3
1
2
2
2
4
3
1
2
3
2
1
2
2
1
3
2
1
2
2
1
3
2
1
2
2
1
3
2
1
2
2
1
1
2
3
0
4
2
1
2
3
0
4
2
1
2
3
0
4
2
1
2
3
0
4
2
3
3
3
5
5
2
3
3
3
5
5
2
3
3
3
5
5
2
3
3
3
5
5
2
4
4
5
3
2
2
4
4
5
3
2
2
4
4
5
3
2
2
4
4
5
3
2
2
(
a)
E
dg
e
quan
t
i
z
at
i
on
(
b)
M1,
s
t
ar
t
i
ng
po
i
nt
(
0,
0)
(
c
)
M2,
s
t
ar
t
i
n
g p
oi
n
t
(
0,
1)
(
d)
M3,
s
t
ar
t
i
ng
po
i
nt
(
1,
0)
1
1
4
3
4
4
1
1
4
3
4
4
0
0
20
40
10
10
0
0
20
40
10
10
2
2
4
3
1
2
2
2
4
3
1
2
30
0
20
40
40
20
30
0
20
40
40
20
3
2
1
2
2
1
3
2
1
2
2
1
30
30
10
30
30
10
30
30
10
30
30
10
1
2
3
0
4
2
1
2
3
0
4
2
10
20
0
0
30
20
10
20
0
0
30
20
3
3
3
5
5
2
3
3
3
5
5
2
20
60
60
40
40
10
20
60
60
40
40
10
4
4
5
3
2
2
4
4
5
3
2
2
10
10
20
30
20
10
10
1
0
20
30
20
10
(
e)
M4,
s
t
ar
t
i
ng
poi
nt
(
1,
1)
(
f
)
E
dge m
ap
(
M1+
M2
+
M3
+
M4)
(
g)
E
dg
e m
ap on
c
ol
or
qu
ant
i
z
at
i
on
(
h)
C
ol
or
quan
t
i
z
at
i
on
0
0
20
40
10
10
1
1
4
3
4
4
1
1
4
3
20
30
0
20
40
40
20
2
2
4
3
1
2
2
2
4
3
1
30
0
20
40
20
30
30
10
30
30
10
3
2
1
2
2
1
3
2
2
2
30
30
30
10
10
20
0
0
30
20
1
2
3
0
4
2
4
20
0
20
20
60
60
40
40
10
3
3
3
5
5
2
3
3
5
5
2
20
60
60
40
40
10
10
10
20
30
20
10
4
4
5
3
2
2
4
4
2
20
20
1
0
(
i
)
C
ol
or
m
ap
(
M1+
M2
+
M3
+
M4)
(
j
)
C
ol
or
m
ap on
edge
qu
ant
i
z
at
i
on
(
k
) M
i
c
ro
-
s
t
r
uc
t
ur
e
f
eat
ur
e of
qua
nt
i
z
e
d
H
S
V
c
ol
or
(l
) M
i
c
ro
-
s
t
r
uc
t
ur
e
f
eat
ur
e of
qua
nt
i
z
e
d
edge
or
i
ent
a
t
i
o
n
F
i
gur
e 3.
P
r
oc
es
s
of
f
i
ndi
ng
m
i
c
r
o
-
s
t
r
uc
t
ur
e
m
ap and m
i
c
r
o
-
s
t
r
uc
t
ur
e
f
eat
ur
e
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
I
mag
e R
et
r
i
ev
a
l
B
as
ed
on
Mul
t
i
S
t
r
uc
t
ur
e
Co
-
o
c
c
ur
r
en
c
e
D
es
c
r
i
pt
or
(
A
gus
Ek
o
Mi
nar
no
)
1179
A
f
t
e
r m
i
c
ro
-
s
t
r
uc
t
ur
e m
ap
(
M
)
i
s
obt
a
i
ne
d,
t
h
e nex
t
pr
o
c
es
s
i
s
f
i
ndi
n
g t
he m
i
c
r
o
-
s
t
r
uc
t
ur
e
f
eat
ur
e bas
ed on
M
,
i
n bot
h
edge or
i
ent
at
i
on i
m
age t
hat
has
bee
n qua
nt
i
z
e
d an
d H
S
V
i
m
age
t
hat
h
as
bee
n qu
ant
i
z
ed.
M
i
s
as
a m
as
k
.
I
f
w
e
w
a
nt
t
o
f
i
nd m
i
c
r
o
-
s
t
r
uc
t
ur
e f
eat
ur
e i
n
qua
nt
i
z
e
d
H
S
V
i
m
age,
M
obt
a
i
n
ed
f
r
om
quant
i
z
ed
edg
e
d
et
ec
t
i
on
i
m
age
i
s
s
up
er
i
m
pos
ed
ov
er
qua
nt
i
z
e
d
H
S
V
i
m
age,
as
i
l
l
us
t
r
at
e
d i
n F
i
g
ur
e 3
(
a)
-
3
(
g
)
.
(
a)
(
b)
F
i
gur
e 4.
F
e
at
ur
e r
ep
r
es
ent
at
i
o
n of
M
S
C
D
:
(
a)
B
at
i
k
C
l
ot
h I
m
age.
(
b)
F
e
at
ur
e
ex
t
r
ac
t
i
on
r
es
ul
t
D
o
t
he
s
am
e
w
a
y
t
o
f
i
nd
m
i
c
r
o
-
s
t
r
uc
t
ur
e
f
eat
ur
e
i
n
q
uant
i
z
ed
e
dge
d
et
ec
t
i
on
i
m
age,
M
obt
a
i
ne
d
f
r
o
m
quant
i
z
ed
H
S
V
i
m
age
i
s
s
uper
i
m
pos
ed
ov
er
qua
nt
i
z
e
d
edg
e
det
ec
t
i
on
i
m
age,
as
i
l
l
us
t
r
at
e
d i
n F
i
gur
e
3
(
h)
-
3
(j
)
.
M
ic
r
o
-
s
t
r
uc
t
ur
e f
eat
u
r
es
ar
e obt
a
i
n
ed f
r
o
m
v
al
u
es
of
pi
x
el
s
t
hat
t
hei
r
pos
i
t
i
on
i
s
al
i
g
ned
w
i
t
h
M
.
W
hi
l
e
v
a
l
ues
of
pi
x
el
s
t
hat
t
hei
r
pos
i
t
i
on
i
s
not
al
i
gne
d
w
i
t
h
M,
ar
e neg
l
ec
t
e
d.
F
i
g
ur
e 3(
k
)
and 3(
l
)
ar
e m
ic
r
o
-
s
t
r
uc
t
ur
e f
eat
ur
e of
quant
i
z
ed
H
S
V
c
o
l
or
an
d
m
i
cr
o
-
s
t
r
uc
t
ur
e f
ea
t
ur
e of
quan
t
i
z
ed
ed
ge
or
i
en
t
at
i
on
r
es
pec
t
i
v
el
y
.
T
hen
,
t
h
os
e Mi
c
r
o
-
s
t
r
uc
t
ur
e
f
eat
ur
es
ar
e
pl
o
t
t
ed
t
o
hi
s
t
ogr
am
as
s
how
n
i
n
F
i
gur
e
4
.
F
i
gur
e
4
(
b)
i
s
f
eat
ur
e
ex
t
r
ac
t
i
on
r
es
ul
t
us
i
ng
MS
C
D
of
B
at
i
k
i
m
age i
n F
i
gur
e
4
(a
).
4.
G
r
a
y
L
ev
el
C
o
-
o
ccu
r
r
en
c
e M
at
r
i
x
T
he us
e of
G
r
a
y
Le
v
e
l
C
o
-
oc
c
ur
r
enc
e M
at
r
i
x
(
G
L
C
M)
i
s
per
f
or
m
ed t
o
det
e
c
t
f
our
f
eat
ur
es
,
t
hes
e
ar
e
en
er
g
y
,
ent
r
op
y
,
c
ont
r
as
t
an
d
c
or
r
el
at
i
o
n
i
n
f
our
di
r
ec
t
i
ons
;
0
°,
45°,
9
0°
a
nd
135°
;
s
o t
her
e ar
e 1
6 f
eat
ur
es
t
ot
al
l
y
.
T
he f
i
r
s
t
s
t
ep
of
G
LC
M i
s
t
r
ans
f
or
m
i
ng t
he R
G
B
i
m
age
i
nt
o
gr
a
y
s
c
al
e
i
m
age.
T
he s
ec
ond
s
t
ep
i
s
c
r
eat
i
n
g
a c
o
-
oc
c
ur
r
enc
e m
at
r
i
x
.
T
he t
hi
r
d
s
t
ep
i
s
dec
i
d
i
ng
a
s
pat
i
al
r
el
at
i
ons
hi
p
bet
w
ee
n t
he r
ef
er
enc
e
pi
x
el
an
d t
he
ne
i
gh
bour
pi
x
el
.
P
ar
am
et
er
s
t
hat
ar
e c
ons
i
der
e
d ar
e
e
dge
or
i
e
nt
at
i
on
(θ
)
an
d d
i
s
t
anc
e
(
d)
.
F
ur
t
her
m
or
e,
t
he n
ex
t
s
t
ep
i
s
c
r
eat
i
ng
a
s
y
m
m
et
r
i
c
al
m
at
r
i
x
b
y
add
i
ng
c
o
-
oc
c
ur
r
enc
e
m
at
r
i
x
w
i
t
h
i
t
s
t
r
ans
pos
e
d
m
at
r
i
x
.
T
hen,
i
t
i
s
f
ol
l
o
w
e
d
b
y
nor
m
al
i
z
i
n
g
t
he
s
y
m
m
et
r
i
c
al
m
at
r
i
x
by
c
om
put
i
ng
t
he
pr
o
bab
i
l
i
t
y
of
eac
h
m
at
r
i
x
el
em
ent
.
Mor
e
ov
er
,
t
he f
i
n
al
s
t
ep i
s
c
om
put
i
ng G
LC
M f
eat
ur
es
.
E
ac
h f
eat
ur
e i
s
c
o
m
put
ed b
y
one
di
s
t
a
nc
e
p
i
x
el
i
n
f
our
di
r
ec
t
i
ons
,
t
h
os
e
ar
e
0°,
45
°,
90
°
an
d
1
35°,
t
o
det
ec
t
c
o
-
oc
c
ur
r
enc
e.
W
h
en G
LC
M has
a m
at
r
i
x
w
i
t
h a s
i
z
e
Lx
L
,
i
n
w
h
i
c
h
L
i
s
t
he num
ber
of
gr
a
y
l
e
v
e
l
s
of
t
he or
i
gi
na
l
im
a
ge
and
w
he
n
t
h
e
pr
o
ba
bi
l
i
t
y
of
pi
x
e
l
i
i
s
t
he
n
ei
gh
bour
of
pi
x
el
j
w
it
h
in
d
is
t
a
n
c
e
d
an
d
e
dge
or
i
ent
at
i
on
θ
is
P
,
t
h
e
e
ner
g
y
f
ea
t
ur
e,
t
he
ent
r
o
p
y
f
e
at
ur
e,
t
he
c
ont
r
as
t
f
e
at
ur
e
a
nd
t
he
c
or
r
el
at
i
on f
eat
ur
e c
an
be c
al
c
ul
at
e
d b
y
e
quat
i
o
ns
(
9)
,
(
1
0
),
(1
1
)
a
nd (
1
2
).
=
∑
2
(
,
,
,
)
−
1
,
=
0
(
9)
=
∑
(
,
,
,
)
.
l
og
(
,
,
,
)
−
1
,
=
0
(
10)
=
∑
(
−
)
2
.
(
,
,
,
)
−
1
,
=
0
(
11)
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
:
11
75
–
1
182
1180
=
∑
(
−
)
−
(
,
,
,
)
−
1
,
=
0
(
12)
W
he
r
e
=
∑
.
(
,
,
,
)
−
1
,
=
0
,
=
∑
.
(
,
,
,
)
−
1
,
=
0
,
=
∑
(
−
)
2
.
(
,
,
,
)
−
1
,
=
0
,
=
∑
(
−
)
2
.
(
,
,
,
)
−
1
,
=
0
.
E
n
er
g
y
,
a
l
s
o c
a
l
l
e
d as
A
n
gul
ar
S
ec
ond
M
om
ent
,
i
s
t
h
e r
epr
es
e
nt
at
i
on
of
i
m
age
hom
ogene
i
t
y
.
W
h
en
t
he
v
a
l
u
e
of
ener
g
y
i
s
h
i
gh,
t
he
r
e
l
at
i
ons
h
i
ps
be
t
w
een
p
i
x
el
s
ar
e
hi
g
hl
y
hom
ogeno
us
.
E
nt
r
o
p
y
i
s
t
h
e opp
os
i
t
e
of
ener
g
y
w
h
i
c
h r
epr
es
ent
t
he
r
an
d
om
nes
s
v
al
ue
bet
w
e
en i
m
ages
.
A
hi
g
her
v
a
l
ue of
ent
r
o
p
y
i
n
di
c
at
es
t
hat
t
he r
e
l
at
i
o
ns
bet
w
ee
n p
i
x
el
s
ar
e
h
i
gh
l
y
r
andom
.
C
ont
r
as
t
i
s
a
v
ar
i
a
t
i
on
of
an
i
m
age’
s
gr
a
y
l
ev
el
.
A
n
i
m
age
w
i
t
h
a
s
m
oot
h
t
ex
t
ur
e
has
a
l
o
w
c
ont
r
as
t
v
a
l
u
e a
nd
an
i
m
age w
i
t
h
a r
o
ugh
t
ex
t
ur
e
has
a
hi
gh
c
ont
r
as
t
v
al
u
e.
C
or
r
el
at
i
o
n
i
s
a
lin
e
a
r
r
e
la
t
i
o
ns
hi
p
b
et
w
e
en
pi
x
e
l
s
.
L
et
,
eac
h
G
LC
M
f
eat
ur
e
0°,
45
°
,
9
0°
an
d
1
35
°
as
H
as
m
(θ
),
H
e
n
t
(θ
),
H
c
o
n
t
(θ
),
H
c
o
rr(θ
)
w
her
e
=
∈
0
,
4
5
,
9
0
,
1
3
5
.
S
o,
t
he f
eat
ur
e
of
G
LC
M i
s
den
ot
e
d as
f
o
llo
w
s
:
=
[
(
0
°
)
…
(
1
3
5
°
)
,
(
0
°
)
…
(
1
3
5
°
)
,
(
0
°
)
…
(
1
3
5
°
)
,
(
0
°
)
.
.
.
(
1
3
5
°
)
]
(
13)
S
o,
t
h
e f
i
nal
f
eat
ur
e of
a bat
i
k
i
m
age i
s
a c
om
bi
nat
i
o
n of
E
MS
D
f
eat
ur
es
and
G
LC
M f
eat
ur
es
deno
t
ed
as
f
ol
l
o
w
s
:
=
[
,
]
(
14)
F
or
ex
a
m
pl
e,
i
f
w
e s
et
t
h
e quant
i
z
at
i
o
n of
c
ol
or
=
72,
i
t
c
ons
i
s
t
s
of
R
=
8,
G
=
3,
B
=
3;
quan
t
i
z
at
i
on
ed
ge
or
i
en
t
at
i
on
=
6
and
t
he
t
ot
a
l
G
LC
M
f
eat
ur
es
ar
e 1
6,
s
o t
he t
ot
a
l
f
eat
ur
e =
94.
5.
P
er
f
o
r
m
an
c
e M
easu
r
e
F
or
eac
h t
em
pl
at
e
i
m
a
ge i
n t
he d
at
as
et
,
an
K
-
di
m
ens
i
ona
l
f
eat
ur
e
v
ec
t
or
T
=
[T
1
,T
2
,…
T
M
]
is
e
x
t
r
ac
t
e
d and s
t
or
e
d i
n t
he da
t
ab
as
e.
Let
Q=
[
Q1
,
Q2
,
…
Q
M
]
be t
he f
eat
ur
e
v
ec
t
or
of
a
quer
y
i
m
age a
n
d t
he
di
s
t
anc
e b
et
w
e
en t
he
m
i
s
s
i
m
pl
y
c
a
l
c
ul
at
ed
as
:
(
,
)
=
∑
|
−
|
|
+
|
+
|
+
|
=
1
(
15)
W
h
er
e
=
∑
=
1
and
=
∑
=
1
.
T
he c
l
as
s
l
a
bel
s
of
t
h
e t
em
pl
at
e i
m
age t
ha
t
y
i
e
l
d
t
he
s
m
al
l
es
t
di
s
t
anc
e
w
i
l
l
be
a
s
s
i
gned t
o t
he
quer
y
i
m
age.
I
n t
h
i
s
ex
per
i
m
ent
,
p
er
f
or
m
anc
e w
as
m
eas
ur
ed us
i
ng pr
ec
i
s
i
o
n a
nd r
ec
al
l
w
hi
c
h
ar
e d
ef
i
ned
as
f
ol
l
o
w
s
:
=
_
_
* 1
0
0
(
16)
=
_
_
* 1
0
0
(
17)
Wh
er
e
i
mg_r
et
r
i
ev
ed
i
s
t
he
num
ber
of
r
et
r
i
ev
ed
i
m
age
s
,
i
mg_r
e
l
ev
a
nt
i
s
t
h
e
num
ber
of
r
el
ev
ant
r
et
r
i
e
v
ed
i
m
ages
,
and
i
mg_
dat
a
bas
e
i
s
t
he
num
ber
of
al
l
r
el
ev
ant
i
m
ages
i
n t
he d
at
ab
as
e.
6.
R
esu
l
t
s
an
d
D
i
s
cu
ssi
o
n
I
n t
h
i
s
s
t
ud
y
,
t
h
er
e ar
e
4 f
eat
ur
e ex
t
r
ac
t
i
on
m
et
h
ods
of
C
B
I
R
us
ed t
o t
e
s
t
t
he
per
f
or
m
anc
e of
MS
C
D
,
na
m
el
y
,
MS
D
,
E
MS
D
,
G
LC
M,
and
MS
C
D
.
M
S
D
has
7
2 f
eat
ur
es
,
E
M
S
D
has
78 f
eat
ur
es
,
an
d G
LC
M has
16
f
eat
ur
es
,
w
h
er
ea
s
MS
C
D
has
94 f
eat
ur
es
.
W
e
c
o
m
par
e t
he
per
f
or
m
anc
e am
ong t
hos
e
m
et
hods
i
n
or
d
er
t
o
k
now
t
he
p
er
f
o
r
m
anc
e MS
C
D
aga
i
ns
t
t
o
i
t
s
pr
edec
es
s
or
(
M
S
D
a
nd
E
M
S
D
)
an
d t
h
e f
am
ous
G
LC
M.
A
n
i
m
age of
eac
h c
l
as
s
on
dat
as
et
i
s
t
ak
en r
and
om
l
y
t
o be
us
ed
as
t
es
t
i
n
g d
at
as
e
t
,
t
hus
w
e
ha
v
e
50
t
es
t
i
n
g
i
m
age
dat
as
et
s
.
E
ac
h
t
es
t
i
n
g
i
m
age
w
i
l
l
be
t
es
t
t
o
r
et
r
i
e
v
e
s
om
e
i
m
age
t
ha
t
hav
e
s
i
m
i
l
ar
c
har
ac
t
er
i
s
t
i
c
w
i
t
h
i
t
’s
f
r
o
m
t
he
r
e
m
ai
n
i
m
ages
on
d
at
as
et
.
T
her
e
ar
e
3
s
c
he
m
es
t
o
k
now
t
h
e
p
er
f
or
m
anc
e
r
es
ul
t
b
as
ed
on
t
he
num
ber
of
i
m
ages
t
hat
m
us
t
be
r
et
r
i
ev
ed,
n
am
el
y
4
,
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
I
mag
e R
et
r
i
ev
a
l
B
as
ed
on
Mul
t
i
S
t
r
uc
t
ur
e
Co
-
o
c
c
ur
r
en
c
e
D
es
c
r
i
pt
or
(
A
gus
Ek
o
Mi
nar
no
)
1181
6,
an
d 8 r
et
r
i
e
v
a
l
i
m
ages
.
S
c
hem
e of
4 r
et
r
i
ev
al
i
m
ages
i
s
c
hos
e
n t
o k
no
w
t
he m
ax
i
m
u
m
pr
ec
i
s
i
o
n
t
h
at
c
a
n
ac
h
i
e
v
e.
T
hen,
s
c
hem
e
of
6
r
et
r
i
e
v
al
i
m
ages
i
s
c
hos
en
as
a
b
as
el
i
ne
of
t
hi
s
C
B
I
R
,
b
ec
aus
e t
h
er
e ar
e
6 i
m
ages
i
n eac
h c
l
as
s
of
dat
as
et
.
W
her
eas
,
s
c
he
m
e of
8 r
et
r
i
ev
a
l
i
m
ages
i
s
c
hos
en t
o k
now
t
he m
ax
i
m
u
m
r
ec
al
l
t
hat
c
an
ac
hi
e
v
e
.
T
abl
e 1 a
nd T
abl
e
2 s
ho
w
t
he per
f
or
m
anc
e r
es
ul
t
s
of
f
our
f
eat
ur
e ex
t
r
ac
t
i
on m
et
h
ods
of
C
B
I
R
.
A
c
c
or
di
n
g t
o t
h
at
t
a
bl
e,
MS
C
D
i
s
s
uper
i
or
t
o
MS
D
,
E
M
S
D
,
a
nd G
LC
M i
n al
l
s
c
hem
es
.
A
dd
i
n
g s
om
e f
eat
ur
es
t
o
a m
et
hod d
oes
n
ot
gi
v
e
guar
a
nt
y
t
h
at
i
t
c
an
he
l
p
t
he m
et
hod
t
o
ac
hi
e
v
e a
n opt
i
m
al
pr
ec
i
s
i
on or
t
o i
nc
r
eas
e t
h
e pr
ec
i
s
i
on
.
E
v
en s
o,
t
he p
er
f
or
m
anc
e r
es
ul
t
s
s
ho
w
t
hat
our
pr
op
os
ed
m
et
hod f
us
i
ng l
oc
al
f
eat
ur
es
and g
l
ob
al
f
eat
ur
es
c
a
n i
nc
r
eas
e t
h
e
pr
ec
i
s
i
o
n ef
f
ec
t
i
v
e
l
y
an
d s
i
gn
i
f
i
c
ant
l
y
.
T
her
e
b
y
i
t
i
n
d
i
c
at
e
s
t
h
at
f
eat
ur
e ex
t
r
ac
t
i
on r
es
u
l
t
s
of
MS
C
D
i
s
m
or
e v
i
s
ual
l
y
r
e
l
ev
a
nt
c
om
par
ed
t
o MS
D
,
E
M
S
D
,
an
d G
LC
M.
T
he r
es
ul
t
s
s
ho
w
t
hat
add
i
ng
G
LC
M
f
eat
ur
es
on
E
M
S
D
i
s
t
he
r
i
ght
dec
i
s
i
on
bec
aus
e
i
t
c
an
i
nc
r
eas
e
t
he
pr
ec
i
s
i
on
of
C
B
I
R
.
G
LC
M
i
s
a
w
el
l
-
k
n
o
w
n f
eat
ur
e
ex
t
r
ac
t
i
on m
et
hod f
or
i
t
s
r
e
l
i
ab
l
e a
nd s
i
m
pl
e enou
gh.
G
LC
M r
epr
es
e
nt
s
f
eat
ur
es
gl
o
bal
l
y
,
w
h
er
eas
E
MS
D
r
e
pr
es
ent
s
f
eat
ur
es
l
oc
a
l
l
y
.
T
hus
,
MS
C
D
i
s
a c
om
bi
nat
i
on m
et
hod
of
l
o
c
al
m
et
hod a
nd g
l
o
bal
m
et
hod.
B
as
ed
on
H
ar
al
i
c
k
pr
opos
e
d m
et
hod [
17
]
,
G
LC
M c
al
c
u
l
at
e
d 1
4 f
e
at
ur
es
o
n 4
or
i
e
n
t
at
i
on
ang
l
es
.
W
e
hav
e
d
one s
o
m
e
ex
a
m
i
nat
i
ons
t
o k
now
w
hi
c
h f
eat
ur
es
o
n
w
hi
c
h
or
i
ent
at
i
on
a
ngl
e
s
houl
d b
e us
e
d
i
n
t
hi
s
s
t
ud
y
.
B
ec
aus
e
f
eat
ur
e
r
e
pr
es
ent
at
i
o
n a
nd
or
i
e
nt
at
i
on
a
ng
l
e
ar
e
i
m
por
t
ant
t
hi
ngs
i
n
G
LC
M
t
hat
c
an
i
nf
l
ue
nc
e
t
he
r
e
t
r
i
e
v
a
l
r
es
ul
t
of
an
i
m
age.
O
ur
ex
am
i
nat
i
ons
i
nd
i
c
at
e
t
ha
t
e
ner
g
y
,
ent
r
o
p
y
,
c
o
nt
r
as
t
,
a
nd c
or
r
e
l
at
i
o
n
f
eat
ur
es
on
0°,
45
°,
9
0°
,
an
d 13
5°
ar
e t
he
bes
t
c
om
bi
nat
i
on f
eat
ur
es
t
o r
epr
es
e
nt
B
at
i
k
i
m
age,
bec
aus
e t
he
y
c
an i
nc
r
eas
e
t
he pr
ec
i
s
i
on
and r
ec
al
l
s
i
gn
i
f
i
c
ant
l
y
.
W
h
en w
e us
e j
us
t
t
hr
ee or
t
w
o or
one of
t
hos
e or
i
ent
a
t
i
on ang
l
e,
t
h
e
pr
ec
i
s
i
o
n an
d r
ec
a
l
l
ar
e go
i
ng d
o
w
n.
MS
C
D
i
s
s
uper
i
or
c
om
par
ed
t
o
M
S
D
b
y
2
0.
25
%
an
d
1
9.
55%
f
or
av
er
ag
e
pr
ec
i
s
i
o
n
and
av
er
a
ge
r
ec
a
l
l
r
es
p
ec
t
i
v
e
l
y
.
C
om
par
ed
t
o
E
M
S
D
,
MS
C
D
i
s
s
uper
i
or
b
y
19
.
0
3%
a
nd
18
.
33%
f
or
av
er
a
ge pr
ec
i
s
i
on an
d av
er
age r
ec
al
l
r
es
pec
t
i
v
e
l
y
.
Mor
eo
v
er
,
c
om
par
ed t
o G
LC
M,
M
S
CD i
s
s
uper
i
or
b
y
14.
86%
and
14.
67%
f
or
av
er
ag
e pr
ec
i
s
i
on
and
av
er
age
r
ec
al
l
r
es
pec
t
i
v
e
l
y
.
T
abl
e
1
.
A
v
er
ag
e c
om
par
i
s
on of
pr
ec
i
s
i
on
P
er
f
o
r
m
a
n
ce
R
et
r
i
eval
MS
D
EM
S
D
G
L
CM
M
S
C
D
(
P
r
opos
e
d)
P
r
ec
i
s
i
on
(
%
)
4
70.
50
71.
50
78.
00
92.
50
P
r
ec
i
s
i
on
(
%
)
6
61.
67
63.
33
67.
33
84.
67
P
r
ec
i
s
i
on
(
%
)
8
51.
50
52.
50
54.
50
67.
25
A
v
er
age (
%
)
61.
22
62.
44
66.
61
81.
47
T
abl
e
2
.
A
v
er
ag
e c
om
par
i
s
on of
r
ec
al
l
P
er
f
o
r
m
a
n
ce
R
et
r
i
eval
MS
D
EM
S
D
G
L
CM
M
S
C
D
(
P
r
opos
e
d)
R
ec
al
l
(
%
)
4
47.
00
47.
67
52.
00
61.
67
R
ec
al
l
(
%
)
6
61.
67
63.
33
67.
33
84.
67
R
ec
al
l
(
%
)
8
68.
67
70.
00
72.
67
89.
67
A
v
er
age (
%
)
59.
11
60.
33
64
78.
67
7.
C
o
n
c
l
u
s
i
o
n
T
hi
s
s
t
ud
y
ex
p
l
a
i
ns
a c
om
bi
nat
i
o
n m
et
hod
of
l
oc
a
l
f
eat
ur
e ex
t
r
ac
t
i
on m
et
hod a
nd
gl
o
bal
f
eat
ur
e ex
t
r
ac
t
i
on m
et
hod f
or
C
B
I
R
,
nam
el
y
MS
C
D
.
M
S
C
D
c
om
bi
nes
E
M
S
D
an
d G
LC
M as
l
oc
a
l
and
g
l
ob
al
f
eat
ur
e
ex
t
r
a
c
t
i
on
m
et
hod
r
es
pec
t
i
v
e
l
y
.
A
c
c
or
di
ng
t
o
ex
p
er
i
m
ent
a
l
r
es
ul
t
s
,
c
o
m
bi
nat
i
on
of
l
oc
a
l
and
gl
oba
l
f
eat
u
r
e
ex
t
r
ac
t
i
o
n
m
et
hod
c
an
i
m
pr
ov
e
t
h
e
pr
ec
i
s
i
on
ef
f
ec
t
i
v
e
l
y
and s
i
g
ni
f
i
c
ant
l
y
.
S
om
e ex
am
i
nat
i
ons
ha
v
e b
een d
on
e t
o det
er
m
i
ne t
he ap
pr
op
r
i
at
e f
eat
ur
es
t
hat
s
hou
l
d be us
e
d f
or
B
at
i
k
i
m
age r
et
r
i
ev
a
l
,
i
n
or
d
er
t
o
ac
hi
ev
e t
he op
t
i
m
al
pe
r
f
or
m
anc
e.
A
t
G
LC
M,
t
h
e
de
t
er
m
i
nat
i
o
n t
hat
s
hou
l
d
be d
one
i
s
t
he
r
i
ght
s
e
l
ec
t
i
on of
f
eat
ur
e
r
epr
es
ent
at
i
ons
and
or
i
e
nt
at
i
on
an
gl
es
.
W
h
er
eas
,
at
E
MS
D
,
t
h
e
de
t
er
m
i
nat
i
on t
hat
s
h
ou
l
d
be
d
o
ne
i
s
t
he
r
i
ght
quan
t
i
z
at
i
on
s
e
l
ec
t
i
on
on
c
ol
or
s
pac
e
and
e
dge
or
i
ent
at
i
o
n.
T
hos
e
s
el
ec
t
i
ons
ar
e
v
er
y
s
ens
i
t
i
v
e
t
hat
c
a
n i
nf
l
uenc
e
t
he r
es
ul
t
of
C
B
I
R
s
i
gni
f
i
c
a
nt
l
y
,
es
pe
c
i
al
l
y
t
he
pr
ec
i
s
i
on.
A
dd
i
n
g
G
LC
M
f
eat
ur
es
t
o
E
M
S
D
g
i
v
e
a
g
ood
and
s
i
gni
f
i
c
ant
ef
f
ec
t
i
n
MS
C
D
.
I
t
c
an
be
pr
ov
e
n f
r
om
t
he ex
per
i
m
e
nt
al
r
es
ul
t
s
.
MS
C
D
i
s
s
up
er
i
or
c
om
par
ed t
o M
S
D
b
y
20
.
25%
an
d
19.
5
5%
f
or
a
v
er
ag
e
pr
ec
i
s
i
on
a
nd
a
v
er
ag
e r
ec
a
l
l
r
es
pec
t
i
v
e
l
y
.
C
om
par
ed t
o
E
MS
D
,
M
S
CD i
s
s
uper
i
or
b
y
19.
03%
and
18.
3
3%
f
or
av
er
ag
e pr
ec
i
s
i
on
an
d a
v
er
a
ge r
ec
al
l
r
es
pec
t
i
v
e
l
y
.
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
:
11
75
–
1
182
1182
Mor
eo
v
er
,
c
om
par
ed t
o G
L
C
M,
M
S
C
D
i
s
s
uper
i
or
b
y
1
4.
86%
a
nd 1
4.
6
7%
f
or
av
er
age
pr
ec
i
s
i
on
and
av
er
age
r
ec
al
l
r
es
pec
t
i
v
e
l
y
.
R
ef
er
en
ces
[1
]
I
da N
ur
hai
da,
R
ul
i
M
anur
ung,
A
ni
at
i
M
ur
ni
Ary
m
u
rt
h
y
.
P
er
f
or
m
anc
e C
om
par
i
s
on A
n
al
y
s
i
s
F
eat
u
r
e
s
E
x
t
r
ac
t
i
o
n M
et
hods
f
or
B
at
i
k
R
ec
og
ni
t
i
on
. In
Pro
c
.
I
CA
CS
I
S
.
2012.
[2
]
A
H
aak
e.
T
he R
ol
e
of
S
ym
m
e
t
r
y
in
J
av
an
es
e B
at
i
k
P
at
t
e
r
ns
.
J
our
n
al
of
C
om
p
ut
er
s
M
at
h.
A
ppl
i
c
.
1989;
17
(
4
-
6
):
81
5
-
8
26.
[3
]
H
am
i
d
i
n A
S
.
B
a
ti
k
,
W
a
r
i
s
a
n
B
uday
a
A
s
l
i
I
ndon
es
i
a.
Y
ogy
ak
ar
t
a:
P
e
ner
b
i
t
N
A
R
A
S
I
.
2
010.
[4
]
[
V
el
dh
ui
s
en
-
D
j
a
j
a
s
oebr
at
a.
O
n t
he
or
i
g
i
n
and
nat
ur
e of
Lar
angan
:
f
or
bi
dd
en
bat
i
k
p
at
t
er
n
s
f
ro
m
t
he
c
ent
r
al
J
a
v
an
es
e pr
i
nc
i
pa
l
i
t
i
es
.
I
n I
r
ene E
m
er
y
R
ound T
abl
e on
M
us
eum
T
ex
t
i
l
es
,
1979
P
r
oc
.
(
E
el
.
M
.
G
i
tti
n
g
e
r
)
.
19
79:
201
-
221.
[5
]
M
C
heong
,
KS L
o
k
e
.
A
n a
ppr
o
ac
h t
o t
ex
t
ur
e
-
b
as
e
d i
m
age r
e
c
ogn
i
t
i
o
n by
dec
ons
t
r
uc
t
i
ng m
ul
t
i
s
pe
c
t
r
a
l
co
-
o
c
c
ur
r
enc
e m
at
r
i
c
e
s
us
i
ng
T
c
he
bi
c
hef
or
t
ho
gona
l
p
ol
y
no
m
i
al
s
. I
n
P
r
o
c
. IC
P
R
’
0
8
,
I
EEE
.
2008.
[6
]
Ka
r
-
S
engL
ok
e
,
M
ar
c
C
heong
.
E
f
f
i
c
i
ent
T
ex
t
i
l
e
R
e
c
o
gni
t
i
o
n
v
i
a
D
e
c
om
po
s
i
t
i
on
o
f
C
o
-
oc
c
ur
r
en
c
e
M
at
r
i
c
es
.
20
09 I
E
E
E
I
nt
er
nat
i
o
nal
C
on
f
er
e
nc
e
on S
i
gna
l
an
d
I
m
ag
e P
r
oc
es
s
i
n
g A
pp
l
i
c
at
i
on
s
.
20
09
.
[7
]
A H
a
ri
s
R
ang
k
ut
i
,
R
i
z
al
B
r
oer
B
ahaw
er
es
,
A
gu
s
H
ar
j
o
k
o.
B
at
i
k
I
m
age R
et
r
i
e
v
al
B
a
s
ed
on S
i
m
i
l
ar
i
t
y
o
f
S
hape
and T
ex
t
ur
e C
h
ar
ac
t
er
i
s
t
i
c
s
.
In
Pro
c
.
I
CA
CS
I
S
.
2012
.
.
[8
]
J
un Y
ue,
Z
henb
o Li
,
Lu L
i
u,
Z
et
i
an F
u
.
C
ont
en
t
-
ba
s
ed i
m
ag
e r
et
r
i
ev
al
u
s
i
n
g c
ol
or
and t
ex
t
u
r
e
f
u
s
ed
fe
a
tu
r
e
s
.
J
our
nal
of
M
at
hem
at
i
c
al
and C
om
p
ut
er
M
odel
l
i
ng
.
2
011;
54
:
1
121
-
1127
.
[9
]
G
uang
-
H
ai
Li
u,
J
i
ng
Y
u
Y
ang.
I
m
ag
e
R
et
r
i
ev
al
ba
s
ed
on
t
h
e
t
ex
t
on
c
o
-
oc
c
ur
r
e
nc
e
m
a
t
r
i
x
.
J
our
n
al
o
f
P
at
t
er
n R
e
c
o
gni
t
i
on
.
200
8;
41
.
[
10]
G
uang
-
H
ai
L
i
u,
L
ei
Z
ha
ng,
Y
i
ng
-
K
un H
ou
,
Z
uo Y
ong Li
,
J
i
ng Y
u
Y
ang.
I
m
a
ge R
et
r
i
ev
al
bas
ed
o
n
m
u
l
ti
-
t
ex
t
on hi
s
t
ogr
am
.
J
o
ur
na
l
of
P
a
t
t
er
n R
ec
o
gni
t
i
on
.
201
0;
43.
[
11]
G
uang
-
H
ai
Li
u,
Z
uo
-
Y
ong
Li
,
Lei
Z
han
g,
Y
ong X
u.
I
m
a
ge R
et
r
i
ev
al
ba
s
e
d on
m
i
c
r
o
-
s
t
r
uc
t
u
r
e
des
c
r
i
po
r
.
J
our
na
l
of
P
at
t
er
n R
ec
og
ni
t
i
on
.
201
1
.
[
12]
A
gus
E
k
o M
i
nar
no,
N
ani
k
S
u
c
i
at
i
.
B
at
i
k
I
m
ag
e R
et
r
i
ev
al
B
as
ed on C
ol
or
D
i
f
f
er
e
nc
e H
i
s
t
ogr
am
an
d
G
r
ay
Lev
el
C
o
-
O
c
c
ur
e
nc
e M
at
r
i
x
.
T
EL
KO
M
N
I
KA
T
el
ec
om
m
uni
c
at
i
on C
om
put
i
ng E
l
ec
t
r
oni
c
s
an
d
C
ont
r
ol
.
201
4
;
12
(
3
)
:
597
-
60
4
.
[
13]
A
gus
E
ko
M
i
nar
no,
N
ani
k
Su
c
i
a
t
i
.
I
m
age R
et
r
i
ev
al
u
s
i
n
g
M
ul
t
i
T
e
x
t
on C
o
-
O
ccu
r
e
n
ce
D
e
scr
i
p
t
o
r
.
Jo
ur
n
al
o
f
T
h
eor
et
i
c
a
l
an
d A
pp
l
i
ed
I
nf
or
m
at
i
on
T
ec
hno
l
og
y
.
2
014
;
67
(
3
)
.
[
14]
A
gus
Ek
o
M
i
nar
no,
Y
uda
M
unar
k
o,
F
i
t
r
i
Bi
m
a
n
t
o
ro
,
Ari
e
K
ur
ni
aw
ar
dhani
,
N
ani
k
S
uc
ia
t
i
.
T
ex
t
ur
e
F
eat
ur
e
E
x
t
r
ac
t
i
on
us
i
n
g
C
o
-
O
c
c
ur
e
nc
e
M
at
r
i
x
of
S
ub
-
B
and
I
m
ag
e
f
or
B
at
i
k
I
m
ag
e
C
l
as
s
i
f
i
c
at
i
o
n
.
I
C
oI
C
T
201
4
.
B
a
ndun
g.
2
014
.
[
15]
A
gus
E
k
o
M
i
nar
no
,
Y
uda
M
unar
k
o
, F
i
tr
i
B
i
m
a
n
to
r
o
,
Arri
e
K
u
r
ni
aw
ar
dhan
i
,
N
a
ni
k
S
uc
i
at
i
.
B
a
t
i
k
I
m
age
R
et
r
i
ev
al
B
as
ed
on E
n
han
c
e
d M
i
c
r
o S
t
r
u
c
t
ur
e
D
e
s
c
r
i
pt
or
.
A
s
ia
-
P
as
i
f
i
c
C
on
f
er
en
c
e
on
C
om
put
e
r
A
i
ded S
y
s
t
em
E
ngi
neer
i
ng,
A
P
C
A
S
E
.
Ba
l
i
.
2014
:
91
-
96.
[
16]
F
anan
i
A
,
Y
uni
ar
t
i
A
,
S
u
c
i
at
i
N
.
G
eom
e
t
r
i
c
F
ea
t
ur
e
E
x
t
r
ac
t
i
on
of
B
at
i
k
I
m
a
ge
U
s
i
ng
C
ar
d
i
n
a
l
S
p
line
C
ur
v
e
R
epr
es
e
nt
at
i
on.
T
EL
K
O
M
N
I
KA
T
el
ec
om
m
uni
c
at
i
on
C
om
put
i
ng
E
l
ec
t
r
oni
c
s
and
C
ont
r
ol
.
201
4
;
12
(
2
)
:
397
-
40
4
.
[
17]
H
ar
al
i
c
k
,
R
ober
t
M
.
S
t
a
t
i
st
i
ca
l
a
n
d
st
r
uc
t
ur
a
l
a
ppr
oa
c
he
s
t
o t
e
x
t
ur
e
.
P
r
o
c
ee
di
ng
s
of
th
e
I
E
EE 6
7
.
1979;
5
:
78
6
-
8
04.
[
18]
S
e
ty
a
w
a
n
I,
T
i
m
o
ti
u
s
IK
,
K
al
v
i
n M
.
A
ut
om
at
i
c
bat
i
k
m
ot
i
f
s
c
l
as
s
i
f
i
c
at
i
on
us
i
ng
v
ar
i
ou
s
c
om
bi
nat
i
on
s
of
S
I
F
T
f
e
at
ur
e
s
m
om
ent
s
an
d
k
-
N
ear
es
t
N
ei
gh
bor
.
7t
h
I
n
t
er
nat
i
ona
l
C
on
f
er
en
c
e
on
I
nf
or
m
at
i
o
n
T
ec
hnol
o
gy
and E
l
ec
t
r
i
c
al
E
ng
i
ne
e
r
i
n
g
, IC
IT
E
E
:
IE
E
E
.
201
5
:
269
-
27
4.
Evaluation Warning : The document was created with Spire.PDF for Python.