I
n
d
o
n
e
s
i
a
n
J
o
u
r
n
a
l
o
f
E
l
e
c
tr
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
te
r
S
c
i
e
n
c
e
V
ol
.
8,
N
o.
1,
O
c
t
ob
er
20
17
,
pp
.
107
~
114
D
O
I
:
10.
115
91/
i
j
eec
s
.
v
8
.i
1
.
pp
10
7
-
11
4
1
07
R
ec
ei
v
ed
Ju
l
y
9
,
20
1
7
;
R
ev
i
s
ed
S
ept
em
ber
1
5,
2
01
7
;
A
c
c
ept
ed
S
ept
e
mb
er
3
0
,
2
017
S
ha
pe
Def
ec
t
D
e
te
c
ti
on
us
i
n
g L
oc
a
l
S
ta
nda
r
d D
e
v
i
a
ti
o
n
a
nd
R
ul
e
-
Base
d
Classif
i
er
f
o
r
B
o
t
t
le
Q
u
alit
y
In
s
p
ect
i
o
n
N
o
r
h
a
s
h
i
m
a
h
M
o
h
d
S
aad
*
1
,
N
o
r
N
a
b
i
l
a
h
S
y
a
z
a
n
a
A
b
d
u
l
R
ah
m
an
2
,
A
b
d
u
l
R
a
h
i
m
A
b
d
u
l
l
a
h
3
,
M
o
h
d
J
u
z
a
i
l
a
A
b
d
L
a
ti
f
4
1
,3
,4
C
ent
er
f
or
R
obot
i
c
s
an
d I
n
dus
t
r
i
al
A
ut
o
m
at
i
on,
U
ni
v
e
r
s
it
i
T
ek
ni
k
a
l
M
al
ay
s
i
a M
el
ak
a,
H
a
ng
T
uah
Ja
ya
,
76100
D
ur
i
a
n T
ungga
l
,
M
el
ak
a,
M
al
ay
s
i
a
1,
2
F
ac
ul
t
y
of
E
l
e
c
t
r
on
i
c
and C
o
m
put
er
E
ng
i
neer
i
ng
,
U
n
i
ve
r
si
t
i
T
ek
ni
k
al
M
al
ay
s
i
a M
el
ak
a
,
H
a
ng
T
uah
Ja
ya
,
76100
D
ur
i
a
n T
ungga
l
,
M
el
ak
a,
M
al
ay
s
i
a
3
F
ac
u
l
t
y
of
E
l
ec
t
r
i
c
al
E
n
gi
n
eer
i
ng
,
U
ni
v
er
s
i
t
i
T
ek
ni
k
al
M
al
ay
s
i
a M
el
ak
a,
H
an
g
T
uah
J
ay
a,
76
100 D
ur
i
an
T
unggal
,
M
el
ak
a
,
M
al
ay
s
i
a
4
F
ac
u
l
t
y
of
M
ec
han
i
c
a
l
E
n
gi
ne
er
i
ng
,
U
n
i
ve
r
si
t
i
T
ek
ni
k
al
M
al
a
y
s
i
a M
el
ak
a,
H
a
ng
T
uah
J
ay
a,
7610
0 D
ur
i
an
T
unggal
,
M
el
ak
a
,
M
al
ay
s
i
a
*
C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
m
a
i
l:
nor
ha
s
hi
m
a
h@
ut
e
m
.
ed
u.
m
y
A
b
st
r
act
T
hi
s
paper
pr
e
s
ent
s
s
hap
e
a
nal
y
s
i
s
u
s
i
ng
Lo
c
al
S
t
andar
d
D
ev
i
at
i
on
(
LS
D
)
t
ec
hni
que
t
o d
e
t
ec
t
s
hap
e
def
ec
t
o
f th
e
bot
t
l
e
f
or
pr
odu
c
t
q
ual
i
t
y
i
ns
pec
t
i
on
.
T
he pr
o
pos
ed an
al
y
s
i
s
f
r
a
m
ew
or
k
i
n
c
l
ud
es
s
egm
ent
at
i
o
n,
f
eat
ur
e
ex
t
r
ac
t
i
on
,
and
c
l
as
s
i
f
i
c
a
t
i
on
.
T
he
s
h
ape
of
t
he
bot
t
l
e
w
as
s
egm
e
nt
ed
us
i
ng
LS
D
t
ec
h
ni
qu
e i
n or
der
t
o
obt
a
i
n
hig
h
er
en
han
c
em
ent
a
t
t
he l
ow
c
ont
r
as
t
ar
ea
a
nd l
ow
e
nha
nc
em
ent
at
t
h
e hi
g
h
c
ont
r
as
t
ar
ea.
T
he
c
ont
r
a
s
t
g
ai
n
t
hat
w
as
app
l
i
e
d
i
n
A
da
pt
i
v
e
C
ont
r
as
t
E
nh
anc
em
ent
(
A
C
E
)
al
gor
i
t
hm
,
w
a
s
pr
es
e
nt
ed
i
n
v
er
s
el
y
pr
o
por
t
i
o
nal
t
o
LS
D
i
n
or
d
er
t
o
det
e
c
t
and
e
l
i
m
i
n
at
e
b
ac
k
gr
o
und
n
oi
s
e
at
t
he
bo
t
t
le
edge.
A
f
t
er
t
he
s
egm
ent
at
i
on
pr
oc
e
s
s
,
t
he
par
am
et
er
s
o
f
t
h
e bot
t
l
e s
hap
e s
u
c
h
as
h
ei
g
ht
,
w
i
dt
h,
ar
ea,
and
e
x
te
n
t
w
er
e
e
x
t
r
a
c
t
e
d
an
d
app
l
i
ed
i
n
c
l
a
s
s
i
f
i
c
at
i
o
n
s
t
age
.
T
h
e
r
ul
e
-
ba
s
ed
c
la
s
s
i
f
i
e
r
w
as
u
s
ed
to
c
l
a
s
s
i
f
y
t
he
s
hap
e of
t
he
bot
t
l
e
ei
t
he
r
g
oo
d o
r
def
ec
t
.
T
he
of
f
l
i
n
e
e
x
per
i
m
ent
al
r
es
ul
t
s
e
x
hi
bi
t
s
up
er
i
or
s
egm
ent
at
i
on
on
per
f
or
m
an
c
e
w
i
t
h 100
%
a
c
c
ur
ac
y
f
or
100
s
am
pl
e
i
m
age
s
.
T
hi
s
s
h
ow
s
t
hat
t
he
LS
D
c
ou
l
d
be a
n ef
f
ec
t
i
v
e
t
ec
h
ni
qu
e t
o m
o
ni
t
or
t
he
pr
od
u
c
t
qu
al
i
t
y
.
Ke
y
w
o
rd
s
:
S
h
ap
e
de
f
ec
t
de
t
ec
t
i
on
;
Lo
c
a
l
S
t
an
dar
d D
ev
i
at
i
o
n;
A
dap
t
i
v
e C
ont
r
as
t
E
nha
nc
e
m
ent
C
o
p
y
r
i
g
h
t
©
2
01
7
I
n
s
t
i
t
u
t
e
o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
r
i
n
g
a
n
d
S
c
i
e
n
c
e
.
A
l
l
r
i
g
h
t
s r
es
er
ved
.
1
.
I
n
tr
o
d
u
c
ti
o
n
I
n t
he
hi
s
t
or
y
of
de
v
e
l
opm
ent
an
d gr
o
w
t
h ec
o
nom
i
c
s
,
v
i
s
ual
i
ns
p
ec
t
i
o
n
is
w
id
e
l
y
u
s
e
d
to
m
oni
t
or
t
he
pr
o
duc
t
q
ual
i
t
y
and
r
e
l
i
ab
i
l
i
t
y
i
n
m
anuf
ac
t
ur
i
ng
i
n
dus
t
r
i
es
[
1]
.
T
he
au
t
om
at
i
c
v
i
s
ua
l
i
ns
pec
t
i
on
i
s
of
t
en bei
ng ut
i
l
i
z
e
b
y
t
h
e i
n
dus
t
r
y
b
ec
aus
e i
t
pr
o
v
i
de
s
hi
gh ef
f
i
c
i
enc
y
and ac
c
ur
ac
y
dur
i
n
g pr
od
uc
t
i
o
n and
i
ns
pec
t
i
o
n pr
oc
es
s
es
[
2]
.
Ma
n
y
c
om
m
er
c
i
al
pr
od
uc
t
s
s
uc
h as
t
ex
t
i
l
e,
f
ood,
be
v
er
age,
an
d m
edi
c
i
ne
em
pl
o
y
e
d
v
i
s
u
al
i
ns
p
ec
t
i
on
t
o
m
ai
nt
ai
n t
hei
r
pr
oduc
t
qua
l
i
t
y
.
P
r
oduc
t
i
ns
pec
t
i
on i
n
t
he
i
nd
us
t
r
y
i
s
i
m
por
t
ant
bec
aus
e
po
or
pr
od
uc
t
qua
l
i
t
y
w
il
l
i
nf
l
uenc
e
c
us
t
om
er
l
o
y
al
t
y
and
i
t
i
s
n
ot
go
od f
or
lo
n
g
-
t
er
m
m
ar
k
et
i
n
g [
3
,
4]
.
T
he m
anual
i
ns
p
ec
t
i
on
i
n
m
os
t
m
anuf
ac
t
ur
i
ng
pr
oc
es
s
m
ai
nl
y
dep
ends
on
h
um
an
oper
at
or
t
o i
ns
p
ec
t
and m
oni
t
or
pr
oduc
t
q
ual
i
t
y
.
H
o
w
ev
er
,
t
hi
s
t
r
ad
i
t
i
ona
l
m
et
hod i
s
pr
on
e
to
hum
an
er
r
or
dur
i
ng
t
he
i
ns
pec
t
i
o
n
pr
oc
es
s
due
t
o
f
at
i
gu
e,
i
l
l
nes
s
,
d
ul
l
a
nd
c
ar
el
es
s
[
5
]
.
G
ener
al
l
y
,
t
h
e
S
m
al
l
a
nd
Med
i
um
E
nt
er
pr
i
s
e (
S
M
E
)
c
o
m
pani
es
w
er
e f
ul
l
y
ut
i
l
i
z
ed t
h
e
m
anual
i
ns
pec
t
i
on
i
n t
he
pr
od
uc
t
i
o
n
li
n
e
t
o
pr
od
uc
e
pr
oduc
t
s
at
l
o
w
co
st
i
n
s
hor
t
per
i
od of
m
anuf
ac
t
ur
i
ng
pr
oc
es
s
[6
].
T
her
ef
or
e,
an
ef
f
ec
t
i
v
e s
ha
pe de
t
ec
t
i
on
t
ec
hni
que
us
i
ng
i
m
age pr
oc
es
s
i
ng
c
oul
d
enha
nc
e
t
h
e
m
anuf
ac
t
ur
i
n
g
pr
oc
es
s
an
d
pr
od
uc
t
qu
a
l
i
t
y
b
ec
aus
e
i
t
pr
ov
i
des
hi
g
h
c
ons
i
s
t
e
nc
y
,
hi
g
h ac
c
ur
ac
y
and
co
st
-
ef
f
e
c
t
i
v
e
nes
s
[7
].
I
n s
hape
det
ec
t
i
o
n t
ec
hn
i
q
u
e,
i
t
i
s
i
m
por
t
ant
t
o c
l
as
s
i
f
y
t
he s
hap
e of
t
he o
bj
ec
t
.
P
r
e
v
i
ous
s
t
udi
es
ha
v
e r
e
por
t
e
d a
c
om
par
i
s
on
bet
w
een
s
hap
e
det
ec
t
i
o
n t
ec
h
ni
que
t
o s
e
l
ec
t
a
pr
op
er
t
ec
hni
que
i
n
or
der
t
o
det
ec
t
s
hape
def
ec
t
.
S
h
ape
det
ec
t
i
on
t
ec
hn
i
qu
es
s
uc
h
as
ada
pt
i
v
e
t
hr
es
hol
di
ng,
m
or
phol
og
i
c
a
l
op
er
at
i
on,
and
e
d
ge
d
et
e
c
t
i
on
w
er
e
us
e
d
t
o
s
e
gm
e
nt
t
he
obj
ec
t
i
n
t
he
i
m
age
[
8]
.
R
o
y
et
al
.
,
(
2014)
di
s
c
o
v
er
e
d
a
dap
t
i
v
e
t
hr
es
hol
di
ng
w
as
not
c
a
pa
bl
e
t
o
s
egm
ent
i
n t
he
h
ig
h
i
l
l
um
i
nat
i
o
n ar
ea of
t
he
i
m
age
and
br
i
g
ht
i
m
age [
9]
.
Mean
w
hi
l
e,
m
or
phol
ogi
c
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
1
07
–
1
14
108
oper
at
i
o
n and e
dge d
et
e
c
t
i
on t
ec
h
ni
q
ues
we
r
e l
e
s
s
ac
c
ur
at
e
and hi
g
h i
n
c
o
m
put
at
i
o
na
l
c
o
m
p
le
x
it
y
f
or
h
ig
h
-
r
es
ol
ut
i
on
i
m
age [
1
0,
11]
.
I
n
a
s
t
ud
y
c
ar
r
i
e
d
o
ut
b
y
C
ve
t
k
o
vi
c
e
t
a
l
.,
(
200
7)
,
L
oc
al
S
t
a
ndar
d
D
e
v
i
at
i
on
(
L
S
D
)
i
s
an
ef
f
ec
t
i
v
e
t
ec
hni
que
t
o
r
edu
c
e
noi
s
e
i
n
t
he
i
m
age
[
12
]
.
T
hus
,
t
he
i
m
age
c
an
be
w
el
l
s
egm
ent
ed
al
t
h
oug
h
t
h
er
e
i
s
an
i
nt
er
r
upt
i
o
n
s
uc
h
as
i
l
l
um
i
nat
i
on
l
ev
e
l
d
ur
i
n
g
i
m
age
ac
q
ui
s
i
t
i
on
.
T
hi
s
pa
per
pr
es
ent
e
d
t
h
e an
al
y
s
i
s
of
s
hape d
ef
ec
t
t
ec
hn
i
qu
e us
i
ng
LS
D
t
o o
bt
a
i
n t
he pr
e
c
i
s
e s
hape f
or
i
m
age
se
gm
ent
at
i
o
n.
LS
D
of
f
er
s
c
ont
r
as
t
en
hanc
em
e
nt
t
ec
h
ni
que
t
o
i
m
pr
ov
e
t
h
e l
o
w
c
o
nt
r
as
t
i
m
age.
I
t
i
s
a
l
s
o
l
o
w
i
n c
om
put
at
i
on
al
c
om
pl
ex
i
t
y
f
or
i
m
age s
egm
ent
at
i
on
c
o
m
par
ed t
o t
h
e
m
or
phol
ogi
c
a
l
o
per
at
i
o
n an
d ed
ge d
et
ec
t
i
on
[
13
].
2.
R
e
sea
r
ch
M
et
h
o
d
T
he i
m
age
w
as
an
al
y
s
ed
u
s
i
ng t
h
e
Mat
l
a
b
s
of
t
w
ar
e i
n
det
ec
t
i
ng t
he d
ef
ec
t
s
of
t
he bot
t
l
e
s
hape.
F
i
gur
e
1
s
ho
w
s
t
he
anal
y
s
i
s
f
r
a
m
ew
or
k
us
i
ng
i
m
age
pr
oc
es
s
i
ng
t
ec
hni
q
u
e.
T
he
s
a
m
pl
e
i
m
age
of
t
he
bot
t
l
e
(
n=
1
00
)
w
as
o
bt
a
i
n
ed
us
i
ng
d
i
g
i
t
a
l
c
a
m
er
a.
T
he
i
m
age
w
as
t
hen
c
on
v
er
t
ed
f
r
o
m
R
G
B
c
ol
or
i
nt
o gr
a
y
s
c
al
e c
ol
or
.
T
he s
hape of
t
he
bot
t
l
e
w
as
s
egm
ent
e
d us
i
ng L
S
D
t
ec
hni
que
t
o
ex
t
r
ac
t
t
he
he
i
ght
,
w
i
dt
h,
ar
ea
,
and
ex
t
e
nt
of
t
he
b
ot
t
l
e.
T
he
r
u
le
-
b
as
ed
c
la
s
s
if
ie
r
w
as
f
i
nal
l
y
i
m
pl
em
ent
ed t
o
c
l
as
s
i
f
y
t
he s
h
ape
def
ec
t
of
t
he
bot
t
l
e.
F
i
gur
e 1.
A
na
l
y
s
i
s
f
r
a
m
ew
o
r
k
2.
1.
Pr
e
-
P
r
o
ces
si
n
g
N
i
k
on D
310
0
di
gi
t
al
c
am
er
a w
i
t
h 14.
2 Me
gap
i
x
e
l
w
as
us
ed t
o c
a
pt
ur
e t
he s
am
pl
e
i
m
age
i
n
R
G
B
c
o
l
or
f
or
m
at
.
I
n
t
he
pr
e
-
pr
oc
es
s
i
n
g
s
t
a
ge,
t
h
e
R
G
B
c
o
l
or
w
as
c
on
v
er
t
ed
i
nt
o
g
r
a
y
sca
l
e
c
ol
or
w
i
t
h
out
c
han
gi
ng t
he
pi
x
el
v
al
ue
[
14
]
.
T
he
br
i
gh
t
nes
s
of
t
he
i
m
age
pi
x
el
w
as
adj
us
t
ed
t
o
be
t
he s
am
e v
al
ue f
or
al
l
pi
x
e
l
’
s
due t
o t
he un
ev
en obj
ec
t
i
l
l
um
i
nat
i
on [
15
]
.
T
he s
m
o
ot
hi
ng pr
oc
es
s
w
as
c
ar
r
i
ed
out
d
ur
i
n
g
t
h
e
pr
e
-
pr
oc
es
s
i
ng
s
t
age
t
o
e
l
i
m
i
nat
e
noi
s
e
an
d
en
hanc
e
t
he
i
m
ag
e
f
or
f
ur
t
her
ana
l
y
s
i
s
[
16]
.
2.
2.
S
e
g
m
e
n
ta
ti
o
n
u
s
i
n
g
L
o
c
a
l
S
t
a
n
d
a
r
d
D
e
v
i
a
ti
o
n
(L
S
D
)
T
he LS
D
t
ec
hn
i
q
ue
w
as
a
ppl
i
ed
i
n
A
da
pt
i
v
e C
ont
r
as
t
E
nh
anc
em
ent
(
A
C
E
)
a
l
g
or
i
t
hm
.
AC
E
a
l
gor
i
t
hm
is
k
now
n
as
an
uns
h
ar
p
m
as
k
i
ng
t
ec
hn
i
qu
e
t
hat
us
ed
f
or
l
o
w
-
f
r
eque
nc
y
c
o
m
ponent
of
t
he
i
m
age
[
1
7]
.
A
C
E
gen
er
al
eq
uat
i
on
i
s
r
epr
es
ent
ed b
y
.
[
]
(,
)
(,
)
(,
)
(,
)
(,
)
x
x
f
i
j
m
i
j
G
i
j
x
i
j
m
i
j
=
+−
(
1)
w
her
e
m
x
(
i
,
j
)
i
s
t
he
l
oc
al
m
ean,
G
(
i
,
j
)
i
s
t
he
c
ont
r
as
t
ga
i
n
an
d
x
(
i
,
j
)
is
g
r
a
y
sca
l
e
v
a
l
u
e
f
or
t
h
e
i
m
age
p
ix
e
l.
T
he
f
unc
t
i
on of
LS
D
us
e
d i
n A
C
E
a
l
g
or
i
t
hm
w
as
t
o c
or
r
ec
t
t
he l
o
w
c
ont
r
as
t
of
an i
m
age.
S
in
g
h
e
t a
l
.,
(
201
2)
obs
er
v
ed i
m
age e
nha
nc
em
ent
w
i
t
h h
i
gh c
o
nt
r
as
t
g
ai
n m
a
y
c
aus
e
r
in
g
in
g
ar
t
i
f
ac
t
s
and
no
i
s
es
t
o
t
he
i
m
age.
T
her
ef
or
e
,
t
he
c
o
nt
r
as
t
ga
i
n
i
n
A
C
E
wa
s
pr
es
e
nt
ed
i
nv
er
s
el
y
pr
opor
t
i
on
al
t
o
LS
D
i
n
or
de
r
t
o
i
nc
r
eas
e
t
he
L
S
D
v
a
l
u
e
and
r
ed
uc
e t
he c
ont
r
as
t
ga
i
n
v
a
l
u
e
w
hi
c
h
det
ec
t
and
e
l
i
m
i
nat
e
b
ac
k
g
r
ound
no
i
s
e
at
t
he
obj
ec
t
edge
[
1
8]
.
T
he us
ed
of
di
f
f
er
ent
ga
i
n w
i
l
l
r
educ
e o
v
er
enh
anc
em
ent
i
n t
he
i
m
age
[1
9
].
T
he L
S
D
i
nf
or
m
at
i
on c
an
be
ex
pr
es
s
ed as
.
[
]
(,
)
(,
)
(,
)
(,
)
(,
)
x
x
x
D
f
i
j
m
i
j
x
i
j
m
i
j
i
j
σ
=
+−
(
2)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
EC
S
IS
S
N
:
2
502
-
4
752
S
ha
pe
D
ef
ec
t
D
et
ec
t
i
on
us
i
ng L
oc
al
S
t
a
ndar
d
D
ev
i
at
i
o
n an
d…
(
N
or
h
as
hi
mah
Mo
hd S
aad
)
109
W
he
r
e
x
(
i
,
j
)
is
t
he
g
r
a
y
sca
l
e
v
al
ue
of
t
he
i
m
age
p
ix
e
l,
f
(
i
,
j
)
i
s
enhanc
e
d v
a
l
ue of
x
(
i
,
j
)
,
m
x
(
i
,
j
)
is
t
he
l
oc
al
m
ean,
σ
x
(
i
,
j
)
i
s
t
he
LS
D
an
d
D
ar
e a
c
ons
t
ant
v
al
u
e.
F
r
om
equa
t
i
on
2,
LS
D
i
s
aut
om
at
i
c
al
l
y
c
ont
r
o
l
l
ed
t
o
s
et
pr
op
er
c
ont
r
as
t
ga
i
n
v
a
l
ue
f
or
t
he
i
m
age.
T
he
f
unc
t
i
on
of
gr
a
y
s
c
al
e
v
al
ue
i
s
s
ho
w
n
i
n
F
i
gur
e
2(
a)
w
h
i
l
e t
h
e f
unc
t
i
on of
LS
D
i
s
i
l
l
us
t
r
at
e
d i
n F
i
gur
e 2(
b)
.
(
a)
(
b)
F
i
gur
e 2.
B
ot
t
l
e i
m
age
(
a)
G
r
a
y
sca
l
e
i
m
age,
(
b)
L
S
D
i
m
age
2.
3.
F
e
a
tu
r
e
E
x
tr
a
c
ti
o
n
T
he
f
eat
ur
e
of
t
he
bot
t
l
e
s
hape
s
uc
h
as
hei
ght
,
w
i
dt
h
,
ar
ea,
and
ex
t
e
nt
we
re
ex
t
r
ac
t
ed
f
r
o
m
t
he bi
nar
y
i
m
age t
o
ob
t
ai
n
t
he
det
ai
l
s
i
nf
or
m
at
i
on.
In
t
h
i
s
s
t
ud
y
,
ex
t
e
nt
p
ar
am
et
er
w
as
u
s
ed
t
o
det
er
m
i
ne
t
he
r
at
i
o
of
t
he
bot
t
l
e
bec
aus
e
of
i
t
s
c
apabi
l
i
t
y
t
o
ob
t
ai
n
a
s
im
ila
r
r
at
i
o
v
al
u
e
of
an
obj
ec
t
al
t
ho
ugh
t
he
i
m
age has
di
f
f
er
ent
p
i
x
el
v
a
l
ue
[
2
0]
,
[
2
1]
.
T
he
ex
t
e
nt
f
or
m
ul
a
i
s
pr
es
e
nt
ed
by
.
=
ℎ
(
3)
T
he s
hape of
t
he
bot
t
l
e
w
a
s
r
epr
es
ent
ed
i
n t
h
e
wh
i
t
e
pi
x
el
as
1 w
hi
l
e b
ac
k
gr
ound i
m
age
w
as
s
h
o
wn
i
n
t
h
e b
l
ac
k
pi
x
el
as
0.
T
he ar
e
a of
t
h
e
o
bj
ec
t
w
as
c
al
c
u
l
at
ed f
r
om
t
he bi
n
ar
y
i
m
age
bas
ed o
n t
h
e v
al
ue of
1 (
w
hi
t
e
pi
x
e
l
)
.
M
ean
w
hi
l
e,
t
he
ar
ea of
t
he
bou
nd
i
ng
box
w
as
det
er
m
i
ned
b
y.
=
ℎ
ℎ
ℎ
(
4)
F
i
gur
e
3 s
ho
w
s
b
ou
ndi
ng
box
di
m
ens
i
on f
or
a
r
ec
t
a
ngl
e
s
hap
e.
T
he
w
i
dt
h
an
d he
i
gh
t
of
t
he
i
m
age w
er
e m
eas
ur
ed
w
i
t
hi
n t
he
r
ang
e
of
x
and
y
ax
es
.
F
i
gur
e 3.
B
ou
nd
i
ng box
d
i
m
ens
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
1
07
–
1
14
110
2.
4.
Ru
l
e
-
b
ased
C
l
a
ssi
f
i
e
r
T
he
r
ul
e
-
bas
e
d
cl
a
ssi
f
i
e
r
w
as
i
m
pl
em
ent
ed
i
n
t
h
i
s
s
t
u
d
y
t
o
c
l
as
s
i
f
y
t
h
e
s
ha
pe
de
f
ec
t
of
t
he
bot
t
l
e
.
T
hi
s
i
s
bec
a
us
e r
ul
e
-
bas
e
d of
f
er
s
s
i
m
pl
e al
gor
i
t
hm
,
eas
y
t
o und
er
s
t
and
an
d hi
g
h
c
l
as
s
i
f
i
c
at
i
on per
f
or
m
anc
e
in
d
et
ec
t
i
ng
t
he d
ef
ec
t
s
bas
ed o
n t
h
e g
i
v
en f
ea
t
ur
es
[
22]
.
In
c
o
m
par
i
ng
w
i
t
h
K
-
n
ear
es
t
nei
ghb
our
(
k
-
N
N
)
c
la
s
s
if
ic
a
t
io
n
,
r
u
le
-
b
as
ed
c
l
as
s
i
f
i
er
c
an
obt
ai
n
h
i
g
h
ac
c
ur
ac
y
f
or
t
he
c
l
as
s
i
f
i
c
at
i
on
s
t
ag
e
an
d
l
o
w
c
om
put
at
i
on
c
om
pl
ex
i
t
y
[
23]
.
W
hi
l
e
k
-
N
N
c
la
s
s
if
ie
r
i
s
hi
gh
c
om
put
at
i
ona
l
c
om
pl
ex
i
t
y
w
he
n
i
nc
r
eas
e
t
h
e
num
ber
of
t
r
ai
ni
n
g
da
t
a
[
24
]
.
T
he
s
hape
of
bot
t
l
e i
s
c
l
as
s
i
f
i
ed as
def
ec
t
c
ondi
t
i
on
w
h
en
t
h
e ex
t
ent
v
a
l
ue i
s
l
es
s
t
han x
v
al
ue o
r
hi
gher
t
h
an
y
v
a
l
ue.
I
f
t
he
ex
t
e
nt
v
a
l
ue
i
s
i
n
b
et
w
e
en
t
he
r
a
nge
of
x
and
y
v
a
l
ue,
t
he
bot
t
l
e
s
h
ape
i
s
i
n
pas
s
c
ondi
t
i
o
n.
T
he pr
oc
es
s
of
c
l
as
s
i
f
i
c
at
i
on
i
s
i
l
l
us
t
r
at
ed
i
n
F
i
gur
e 4.
F
i
gur
e
4.
C
l
as
s
i
f
i
c
at
i
on
pr
o
c
e
s
s
3.
R
e
su
l
t
s
a
n
d
A
n
a
l
y
s
i
s
T
he 100
s
am
pl
es
of
bot
t
l
e
i
m
age,
c
ons
i
s
t
s
of
50
go
od s
h
ape
i
m
ages
an
d
50
s
hap
e
def
ec
t
i
m
ages
w
er
e
an
al
y
s
ed
us
i
ng
i
m
age
pr
oc
es
s
i
ng
t
ec
hn
i
q
ue.
F
i
g
ur
e
5
a
nd F
i
gur
e
6
s
ho
w
t
he ex
am
pl
e
of
t
he
g
ood
an
d def
ec
t
s
hap
e of
t
he
bot
t
l
e
.
F
i
gur
e
5.
E
x
am
pl
es
of
good
s
hape
of
bot
t
l
e
i
m
age
F
i
gur
e
6.
E
x
am
pl
es
of
def
ec
t
s
hape
of
bot
t
l
e i
m
age
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
EC
S
IS
S
N
:
2
502
-
4
752
S
ha
pe
D
ef
ec
t
D
et
ec
t
i
on
us
i
ng L
oc
al
S
t
a
ndar
d
D
ev
i
at
i
o
n an
d…
(
N
or
h
as
hi
mah
Mo
hd S
aad
)
111
F
i
gur
e
7
s
how
s
t
he s
h
ap
e an
al
y
s
i
s
pr
oc
es
s
f
or
t
he
def
ec
t
b
ot
t
l
e
.
T
he
pr
oc
es
s
w
as
s
t
ar
t
ed
w
i
t
h t
he
co
n
ve
r
si
o
n
of
r
ed,
gr
e
en a
nd b
l
u
e
(
R
G
B
)
c
ol
or
i
m
age
i
n F
i
g
ur
e 7(
a)
i
nt
o
a
gr
a
y
s
c
al
e
i
m
age
i
n
F
i
g
ur
e
7(
b)
.
T
he
c
onv
er
s
i
on
pr
oc
e
s
s
w
as
appl
i
e
d
t
o
r
educ
e
t
he
c
om
pl
ex
i
t
y
dur
i
n
g t
he a
na
l
y
s
i
s
.
R
G
B
c
ol
or
i
m
ages
w
er
e
una
bl
e
t
o pr
ov
i
d
e det
ai
l
s
in
f
or
m
at
i
on
i
n or
der
to
i
de
nt
i
f
y
t
he
i
m
por
t
ant
edg
e
s
or
f
eat
ur
es
of
t
h
e
o
bj
ec
t
.
B
es
i
d
es
,
t
he
R
G
B
i
m
age
w
as
eas
y
t
o
be
ex
pos
ed
t
o
t
he
no
i
s
e
v
ar
i
at
i
ons
[
25
]
.
T
hus
,
t
h
e c
o
nv
er
s
i
on
i
nt
o t
he
gr
a
y
s
c
a
l
e i
m
age
w
as
i
m
por
t
ant
i
n r
ed
uc
i
n
g c
om
pl
ex
i
t
y
f
r
om
3 di
m
ens
i
ona
l
(
3D
)
pi
x
el
v
a
l
ue
i
nt
o 2 d
i
m
ens
i
onal
(
2D
)
pi
x
el
v
al
u
e [
26
].
(
a)
(
b)
(c
)
(
d)
F
i
gur
e 7.
S
ha
pe
s
e
gm
ent
at
i
on pr
oc
es
s
T
he
g
r
a
y
sca
l
e
i
m
age
w
as
t
hen
t
r
a
ns
f
or
m
ed
i
nt
o
s
t
a
nd
ar
d
de
v
i
at
i
on
i
m
age
as
s
ho
w
n
i
n
F
i
gur
e
7(
c
)
.
I
n
t
h
i
s
pr
oc
es
s
,
s
t
andar
d
de
v
i
at
i
on
al
gor
i
t
hm
w
as
app
l
i
ed
t
o
s
egm
ent
t
he
s
h
ape
of
t
he bot
t
l
e b
y
s
ep
ar
at
i
ng t
h
e
bo
t
t
l
e
a
nd t
he
bac
k
gr
ound i
m
age.
T
he bac
k
gr
oun
d no
i
s
e
at
t
he
e
dge of
t
he
b
o
ttl
e
c
an b
e
r
em
ov
ed b
y
i
nc
r
eas
i
ng t
he
v
a
l
ue
of
LS
D
r
es
u
l
t
i
ng t
h
e r
educ
t
i
on
of
c
ont
r
as
t
gai
n
of
t
he
i
m
age.
T
her
ef
or
e,
t
he
s
hape
of
t
he
bot
t
l
e
w
as
out
l
i
ned
us
i
ng
t
he
w
hi
t
e
l
i
ne
t
o s
epar
a
t
e b
et
w
een
t
he
s
h
ape
of
t
he
b
o
ttl
e
a
nd t
he b
a
c
k
gr
ound
i
m
age
.
T
he s
t
andar
d dev
i
at
i
o
n i
m
age
w
as
c
onv
er
t
ed
i
nt
o a
bi
n
ar
y
i
m
age as
s
how
n i
n
F
i
gur
e
7(
d)
.
T
he
s
hape
of
t
he
bot
t
l
e
w
as
pr
es
en
t
ed
i
n
w
h
i
t
e
pi
x
el
(
1)
w
h
i
l
e
t
he
bac
k
gr
ound
i
m
age
w
as
i
n t
he
b
la
c
k
pi
x
el
(
0)
.
F
i
n
al
l
y
,
t
he f
eat
ur
e
ex
t
r
ac
t
i
on p
ar
am
et
er
s
w
er
e ex
t
r
ac
t
e
d t
o
be us
e
d f
or
t
he
c
la
s
s
if
ic
a
t
io
n
pr
oc
es
s
w
her
e
ex
t
e
n
t
par
am
et
er
w
as
c
a
l
c
ul
a
t
ed
bas
e
d
on
t
he
r
at
i
o
of
t
he
b
ot
t
l
e
ar
ea
a
nd t
he
bou
ndi
ng
box
ar
ea.
F
our
bot
t
l
e
par
am
et
er
s
of
hei
g
ht
,
w
i
dt
h,
ar
e
a,
a
nd
ex
t
e
nt
ex
t
r
ac
t
ed
f
r
o
m
t
he go
od
and
def
ec
t
s
hape
i
m
ages
w
er
e t
abu
l
at
e
d i
n T
abl
e
1.
A
s
ex
p
ec
t
ed,
bot
h
s
am
pl
e i
m
ages
hav
e s
i
m
i
l
ar
ar
ea of
bo
und
i
n
g box
but
di
f
f
er
ent
v
a
l
u
e of
t
he
bot
t
l
e
ar
ea.
T
he ar
e
a v
al
ue f
or
go
o
d bot
t
l
e
w
as
hi
g
her
c
om
par
ed t
o
def
ec
t
bot
t
l
e
w
hi
c
h t
hen
i
nf
l
u
enc
e
d t
o t
he
ex
t
ent
v
a
l
u
e.
T
abl
e 1.
P
ar
am
et
er
s
of
bot
t
l
e s
ha
pe
S
a
m
pl
e i
m
age
A
r
ea
boundi
ng box
(
pi
x
el
)
A
r
ea
bot
t
l
e
(
pi
x
el
)
E
x
t
ent
G
ood1
1.
29792e+
07
3.
52914e+
06
0.
2719
G
ood2
1.
29792e+
07
3.
50918e+
06
0.
2704
D
ef
ec
t
1
1.
29792e+
07
3.
00041e+
06
0.
2312
D
ef
ec
t
2
1.
29792e+
07
3.
24855e+
06
0.
2503
F
i
gur
e
8 s
h
o
w
s
t
h
e r
ul
e
-
b
as
ed c
l
as
s
i
f
i
er
pr
oc
es
s
t
o
i
de
nt
i
f
y
t
he
def
ec
t
s
ha
pe
of
t
he
bot
t
l
e.
T
he
m
ax
i
m
u
m
and
m
i
ni
m
u
m
of
ex
t
ent
v
al
ues
w
er
e
s
e
l
ec
t
e
d
a
nd
us
ed
i
n
t
he
r
u
le
-
b
as
ed
c
l
as
s
i
f
i
er
al
gor
i
t
hm
.
T
he r
ange
of
t
he
ex
t
ent
v
al
ue
w
a
s
f
ound t
o
be
bet
w
ee
n 0
.
26
12 a
nd
0.
2
761.
F
ro
m
t
h
e
ru
l
e
-
b
as
ed
c
l
as
s
i
f
i
er
,
i
f
t
he ex
t
en
t
v
al
ue
w
as
gr
eat
er
t
ha
n
0
.
27
61 or
l
o
w
er
t
ha
n
0
.
26
12,
t
he s
ha
pe
of
t
he
b
o
ttl
e
w
a
s
c
l
as
s
i
f
i
ed as
a
def
ec
t
c
o
n
d
it
io
n
.
W
hi
l
e i
f
t
he
ex
t
ent
v
a
l
ue
w
as
i
n
bet
w
e
en
0.
26
12
and
0.
2
76
1,
t
h
e s
hap
e of
t
he
bot
t
l
e
w
as
c
l
as
s
i
f
i
ed as
goo
d c
on
di
t
i
on.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
1
07
–
1
14
112
F
i
gur
e
8.
R
ul
e
-
bas
ed c
l
as
s
i
f
i
er
T
he of
f
l
i
ne ac
c
ur
ac
y
of
good a
nd def
ec
t
bot
t
l
e
i
s
i
l
l
us
t
r
at
ed i
n F
i
gur
e 9
.
T
h
e
ex
per
i
m
ent
al
r
es
u
l
t
s
s
ho
w
100%
ac
c
ur
ac
y
f
or
good
a
nd
def
ec
t
b
ot
t
l
es
bas
e
d
on
100 s
am
pl
es
i
m
age.
T
he per
f
or
m
anc
e o
f
t
he anal
y
s
i
s
t
o c
l
as
s
i
f
y
t
h
e bot
t
l
e s
hap
e
def
ec
t
i
s
de
pends
on
t
he
s
a
m
pl
e
num
ber
of
bot
t
l
e
i
m
age.
B
as
ed
on
t
he
p
er
c
ent
ag
e
of
t
he
ac
c
ur
ac
y
,
i
t
s
ho
w
s
t
h
e
pr
opos
e
d t
ec
h
ni
q
ue c
a
pab
l
e
of
det
ec
t
i
ng
t
he
s
hap
e de
f
ec
t
ef
f
ec
t
i
v
el
y
.
F
i
gur
e 9.
A
c
c
ur
ac
y
gr
a
ph
4
.
C
o
n
c
l
u
s
i
o
n
T
hi
s
paper
pr
es
ent
s
t
he
an
al
y
s
i
s
of
s
hape
def
ec
t
d
et
e
c
t
i
on
us
i
ng
L
S
D
te
ch
ni
q
ues
.
T
he
f
i
r
s
t
pr
oc
es
s
i
s
s
t
ar
t
ed
w
i
t
h
a
pr
e
-
pr
oc
es
s
i
ng
s
t
a
ge
w
h
er
e
t
he
i
m
age
i
s
c
or
r
ec
t
ed
and
i
m
pr
ov
ed
t
o
e
lim
in
at
e
t
h
e
n
oi
s
e
i
n
t
h
e
i
m
age.
T
hen,
L
S
D
t
ec
hn
i
que
i
s
us
e
d
t
o
s
egm
ent
t
h
e
s
hap
e
of
t
he
bot
t
l
e.
L
S
D
t
ec
h
ni
que s
h
o
w
s
t
he ab
i
l
i
t
y
t
o s
egm
ent
t
he
s
hape of
t
he
b
o
t
t
le
v
er
y
w
el
l
b
y
r
e
duc
i
ng
c
ont
r
as
t
gai
n
i
n
t
he i
m
age.
N
ex
t
,
p
ar
am
et
er
s
w
hi
c
h
ar
e hei
ght
,
w
i
dt
h
,
ar
e
a,
and
ex
t
ent
par
am
et
er
s
ar
e
ex
t
r
ac
t
ed
f
r
om
t
he
i
m
age
.
E
x
te
n
t
par
a
m
et
er
i
s
us
ed
t
o
c
a
l
c
ul
at
e
t
he
r
at
i
o
of
t
he
bot
t
l
e
ar
ea.
L
as
t
l
y
,
t
he
r
u
le
-
bas
ed
c
l
as
s
i
f
i
er
i
s
app
l
i
ed
t
o
c
l
as
s
i
f
y
t
h
e
def
ec
t
of
t
h
e
s
hape
bo
t
t
l
e
ei
t
h
er
goo
d or
def
ec
t
.
T
he per
f
or
m
anc
e of
t
he
t
ec
hni
q
ue
w
as
v
er
i
f
i
ed
i
n
t
er
m
s
o
f
ac
c
ur
ac
y
.
F
ro
m
th
e
ex
per
i
m
ent
al
r
es
ul
t
,
10
0%
ac
c
ur
ac
y
w
as
ac
hi
ev
e
d
t
o
de
t
ec
t
t
he
s
h
ap
e
d
ef
ec
t
of
t
he
bot
t
l
e
f
or
100
s
am
pl
es
i
m
age.
T
hi
s
i
n
di
c
at
es
t
h
at
t
he
LS
D
t
e
c
hni
q
ue
c
o
ul
d
b
e
a
n
ef
f
ec
t
i
v
e
m
et
hod
t
o
det
ec
t
s
h
ape
def
ec
t
i
n
m
anuf
ac
t
ur
i
ng pr
oc
es
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
EC
S
IS
S
N
:
2
502
-
4
752
S
ha
pe
D
ef
ec
t
D
et
ec
t
i
on
us
i
ng L
oc
al
S
t
a
ndar
d
D
ev
i
at
i
o
n an
d…
(
N
or
h
as
hi
mah
Mo
hd S
aad
)
113
A
c
k
n
o
w
l
e
d
g
e
m
e
n
t
T
he aut
hor
s
w
ou
l
d
l
i
k
e t
o
t
hank
t
he U
n
i
v
er
s
i
t
i
T
ek
ni
k
a
l
Ma
l
a
y
s
i
a
Me
l
ak
a (
U
T
eM)
,
U
T
eM
Z
am
al
ah S
c
hem
e,
R
ehabi
l
i
t
at
i
on E
n
gi
neer
i
ng &
A
s
s
i
s
t
i
v
e T
ec
hnol
og
y
(
R
E
AT
)
r
e
s
ear
c
h gr
oup
under
C
en
t
er
of
R
obot
i
c
s
&
I
nd
us
t
r
i
a
l
A
ut
om
at
i
o
n
(
C
e
R
I
A
)
,
A
d
v
a
nc
ed
D
i
g
i
t
a
l
S
i
g
nal
P
r
oc
es
s
i
ng
(
AD
S
P)
R
es
ear
c
h
L
ab
or
at
or
y
and
Mi
ni
s
t
r
y
of
H
i
gher
E
duc
at
i
on (
MO
H
E
)
,
Mal
a
y
s
i
a f
or
s
pons
or
i
n
g
t
h
i
s
w
or
k
under
pr
oj
ec
t
G
Luar
/
S
T
E
V
I
A
/
20
16/
F
K
E
-
C
eR
I
A
/
l
0
000
9
an
d
t
he
us
e
of
t
he
ex
i
s
t
i
n
g f
ac
i
l
i
t
i
es
t
o c
om
pl
et
e t
h
i
s
pr
oj
ec
t
.
R
ef
er
en
ces
[1
]
Sa
rk
a
r B,
S
a
re
n
S
.
P
r
odu
c
t
i
ns
pe
c
t
i
on p
ol
i
c
y
f
or
an
i
m
per
f
ec
t
pr
odu
c
t
i
on s
y
s
t
em
w
i
t
h i
ns
pe
c
t
i
o
n
er
r
or
s
and w
ar
r
ant
y
c
os
t
.
E
ur
o
pean J
our
nal
of
O
per
at
i
o
nal
R
es
ear
c
h.
201
6
;
2
48(
1)
:
263
-
7
1.
[2
]
G
i
m
enez
T
,
P
i
ov
es
an C
,
B
r
ag
a M
M
,
R
aggi
o D
P
,
D
eer
y
C
,
R
i
c
k
e
t
t
s
D
N
,
E
k
s
t
r
an
d K
R
,
M
ende
s
F
M.
V
i
s
ua
l
i
n
s
pe
c
t
i
on f
or
c
ar
i
e
s
det
e
c
t
i
o
n:
a
s
y
s
t
e
m
at
i
c
r
ev
i
e
w
and m
et
a
-
a
n
a
l
ysi
s
.
J
our
na
l
of
de
nt
a
l
r
es
ear
c
h
.
201
5;
9
4
(7
):
89
5
-
904
.
[3
]
H
uang S
H
,
P
an Y
C
.
A
ut
o
m
at
e
d v
i
s
u
al
i
ns
pec
t
i
on
i
n
t
he s
em
i
c
o
ndu
c
t
or
i
ndu
s
t
r
y
:
A
s
ur
v
ey
.
Co
m
put
er
s
i
n i
ndu
s
t
r
y
.
20
15
;
66
:
1
-
0.
[4
]
J
aha
ns
h
ani
A
A
,
H
a
j
i
z
ade
h G
M
,
M
i
r
dham
ad
i
S
A
,
N
aw
as
er
K
,
K
hak
s
ar
S
M
.
S
t
udy
t
he ef
f
ec
t
s
of
c
us
t
om
er
s
e
r
v
i
c
e and pr
odu
c
t
qual
i
t
y
on c
us
t
om
er
s
at
i
s
f
a
c
t
i
o
n
and l
oy
al
t
y
.
[5
]
M
or
adi
G
,
S
ham
s
i
M
,
S
edaa
ghi
M
H
,
M
or
adi
S
.
A
ppl
e def
ec
t
de
t
ec
t
i
on us
i
ng s
t
at
i
s
t
i
c
a
l
hi
s
t
ogr
a
m
bas
e
d f
uz
z
y
c
-
m
eans
a
l
gor
i
t
h
m
.
I
nM
ac
hi
ne
V
i
s
i
o
n and I
m
a
ge P
r
oc
e
s
s
i
ng (
M
V
I
P
)
.
I
EEE.
2011 7t
h
I
r
ani
a
n 20
11
;
16
;
1
-
5
.
[6
]
K
es
k
i
n
H
,
Ş
e
nt
ür
k
C
,
S
u
ngur
O
,
K
i
r
i
ş
H
m
.
T
he
I
m
por
t
anc
e
o
f
S
M
E
s
I
n
D
ev
el
opi
n
g
E
c
ono
m
i
es
.
[7
]
B
r
os
na
n
T
,
S
un
D
W
.
I
m
pr
ov
i
ng
qual
i
t
y
i
ns
p
ec
t
i
on
of
f
ood
pr
odu
c
t
s
by
c
o
m
put
er
v
i
s
i
o
n
–
a
r
ev
i
ew
.
J
our
n
al
o
f
f
oo
d e
ngi
n
eer
i
ng.
2
004;
6
1
(1
):
3
-
16.
[8
]
P
ai
Y
T
,
C
hang Y
F
,
R
ua
n
S
J
.
A
dapt
i
v
e
t
hr
e
s
ho
l
di
ng
al
gor
i
t
h
m
:
E
f
f
i
c
i
ent
c
o
m
p
ut
at
i
on
t
ec
hn
i
que
ba
s
e
d
on i
n
t
el
l
i
ge
nt
b
l
oc
k
d
et
e
c
t
i
o
n
f
or
de
gr
ad
ed d
oc
u
m
e
nt
i
m
a
ges
.
P
at
t
er
n R
ec
ogn
i
t
i
o
n
.
201
0
;
4
3
(9
):
3177
-
87
.
[9
]
R
oy
P
,
D
ut
t
a S
,
D
ey
N
,
D
ey
G
,
C
hak
r
ab
or
t
y
S
,
R
ay
R
.
A
dapt
i
v
e t
hr
e
s
ho
l
di
ng:
a
c
om
par
a
t
i
v
e
s
t
ud
y
. In
C
ont
r
ol
,
I
n
s
t
r
um
ent
at
i
o
n,
C
o
m
m
uni
c
at
i
on
and
C
om
p
ut
at
i
on
a
l
T
ec
hno
l
og
i
es
(
I
CCI
CCT
).
I
EEE.
2014
I
nt
er
na
t
i
o
nal
C
o
nf
er
e
nc
e on.
2
014
;
1
182
-
1186
.
[
10]
M
ai
ni
R
,
A
ggar
w
al
H
.
S
t
u
dy
and c
o
m
pa
r
i
s
on o
f
v
ar
i
o
us
i
m
ag
e ed
ge d
et
e
c
t
i
o
n t
ec
hn
i
qu
es
.
I
nt
er
na
t
i
o
nal
j
our
nal
o
f
i
m
ag
e
pr
oc
e
s
s
i
ng (
I
J
I
P
)
.
200
9
;
3(
1)
:
1
-
1.
[
11]
P
es
ar
e
s
i
M
,
B
enedi
k
t
s
s
on
J
A
.
A
new
appr
o
ac
h
f
or
t
h
e
m
or
ph
ol
og
i
c
a
l
s
eg
m
ent
at
i
o
n
o
f
h
i
gh
-
r
e
s
o
lu
t
ion
s
at
e
l
l
i
t
e i
m
ag
er
y
.
I
E
E
E
T
r
an
s
a
c
t
i
o
ns
on G
eo
s
c
i
en
c
e a
nd R
e
m
ot
e S
ens
i
ng
.
20
01;
3
9
(2
):
3
09
-
20.
[
12]
C
v
e
t
k
o
v
i
c
SD
,
Sc
h
i
rri
s
J
,
d
e
W
i
t
h
PH
.
L
oc
a
ll
y
-
adap
t
i
v
e
i
m
age
c
on
t
r
as
t
en
han
c
em
ent
w
i
t
hout
no
i
s
e
and r
i
ngi
ng ar
t
i
fa
c
ts
.
In
Im
a
g
e
P
r
o
c
e
s
s
i
n
g
.
I
C
I
P
200
7.
I
E
E
E
I
nt
er
nat
i
on
al
C
o
nf
er
e
nc
e
.
20
07
;
3
: III
-
557
.
[
13]
Y
uguang Y
ang and Z
hen
g Z
hou
.
R
es
ear
c
h and i
m
pl
em
en
t
at
i
o
n
of
i
m
age en
han
c
em
ent
al
gor
i
t
hm
bas
e
d
on
l
o
c
al
m
ean
a
nd
s
t
an
dar
d
dev
i
at
i
on
.
I
E
E
E
S
y
m
pos
i
um
on
E
l
ec
t
r
i
c
al
&
E
l
ec
t
r
oni
c
s
E
ngi
ne
er
i
n
g (
E
E
E
S
Y
M
)
,
K
ual
a
Lum
pur
.
2
012;
375
-
378
.
[
14]
A
t
m
a
ja
R
D
,
M
ur
t
i
M
A
,
H
al
om
oan
J
,
S
ur
at
m
an
F
Y
.
A
n i
m
a
ge
pr
o
c
e
s
s
i
ng m
et
hod
t
o c
on
v
er
t
R
G
B
i
m
ag
e i
nt
o bi
nar
y
.
I
nd
one
s
i
a
n
J
our
n
al
of
E
l
ec
t
r
i
c
al
E
n
gi
n
ee
r
i
ng an
d C
om
put
er
S
c
i
e
nc
e
.
2016
;
3
(2
)
:
377
-
38
2.
[
15]
J
i
an
g G
,
W
o
ng
C
Y
,
Li
n S
C
,
R
ahm
an
M
A
,
R
en T
R
,
K
w
ok
N
,
S
hi
H
,
Y
u Y
H
,
W
u T
.
I
m
a
ge c
ont
r
a
s
t
enhan
c
e
m
ent
w
i
t
h br
i
ght
nes
s
pr
es
er
v
at
i
o
n u
s
i
n
g a
n o
pt
i
m
al
g
am
m
a
c
or
r
e
c
t
i
on
and
w
ei
ght
ed
s
u
m
appr
oa
c
h.
J
o
ur
na
l
of
M
oder
n
O
pt
i
c
s
.
20
15;
62(
7)
:
536
-
47.
[
16]
R
ehm
an A
,
S
ab
a T
.
N
eur
al
net
w
or
k
s
f
or
doc
um
e
nt
i
m
age
pr
epr
o
c
es
s
i
ng:
s
t
at
e of
t
he ar
t
.
A
r
t
if
ic
ia
l
I
nt
el
l
i
ge
nc
e R
ev
i
ew
.
2
014
;
42(
2)
:
253
-
73.
[
17]
P
r
a
k
o
s
o
B
S
,
T
i
m
o
ti
u
s
IK
,
S
e
ty
a
w
a
n
I.
P
al
m
pr
i
nt
i
den
t
i
f
i
c
a
t
i
on
f
or
u
s
er
v
er
i
f
i
c
a
t
i
on
b
as
ed
o
n
l
ine
det
e
c
t
i
o
n
an
d
l
oc
a
l
s
t
an
dar
d
d
ev
i
a
t
i
on
.
I
n
f
or
m
at
i
o
n
T
ec
h
nol
o
gy
,
C
om
pu
t
er
and
E
l
e
c
t
r
i
c
al
E
ngi
ne
er
i
n
g
(I
C
I
T
AC
EE),
I
EEE.
2014 1
s
t
I
n
t
er
nat
i
ona
l
C
onf
er
en
c
e.
201
4;
155
-
15
9.
[
18]
S
i
ngh S
S
,
S
i
n
gh T
T
,
D
ev
i
H
M
,
S
i
nam
T
.
Loc
a
l
c
o
nt
r
as
t
enh
a
nc
e
m
ent
u
s
i
n
g l
o
c
al
s
t
an
dar
d
dev
i
at
i
o
n.
I
nt
er
na
t
i
o
nal
J
our
nal
of
C
om
p
ut
er
A
pp
l
i
c
at
i
ons
.
20
12;
47(
15)
.
[
19]
O
m
er
O
s
am
a
A
,
T
os
hi
h
i
s
a
T
anak
a
.
R
obus
t
i
m
ag
e
r
egi
s
t
r
at
i
o
n
bas
ed
o
n
l
oc
al
s
t
a
ndar
d
de
v
i
a
t
i
on
a
nd
i
m
age i
nt
en
s
i
t
y
.
I
nf
or
m
a
t
i
on
,
C
om
m
u
ni
c
at
i
ons
&
S
i
gna
l
P
r
oc
e
s
s
i
ng,
2
007 6
t
h I
n
t
er
nat
i
ona
l
C
onf
er
en
c
e
on.
I
E
E
E
,
20
07.
[
20]
R
ege
S
,
M
em
ane
R
,
P
hat
ak
M
,
A
gar
w
al
P
.
2D
geom
et
r
i
c
s
hap
e
and
c
ol
or
r
ec
ogn
i
t
i
o
n
u
s
i
ng
di
g
i
t
a
l
i
m
ag
e pr
o
c
es
s
i
n
g.
I
nt
er
nat
i
o
nal
j
o
ur
na
l
of
ad
v
an
c
ed r
e
s
ear
c
h i
n e
l
ec
t
r
i
c
al
,
e
l
ec
t
r
o
ni
c
s
an
d
i
ns
t
r
um
ent
at
i
on eng
i
neer
i
ng.
2
013
;
2(
6)
:
2
479
-
87.
[
21]
P
at
el
S
,
T
r
i
v
edi
P
,
G
an
dhi
V
,
P
r
aj
ap
at
i
G
I
.
2D
b
as
i
c
s
hape d
et
e
c
t
i
o
n u
s
i
ng r
egi
on
pr
oper
t
i
es
.
I
nt
er
na
t
i
o
nal
J
our
nal
of
E
n
gi
n
e
er
i
ng R
e
s
e
ar
c
h
&
T
e
c
hno
l
og
y
.
2013;
2
(5
):
114
7
-
53.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
1
07
–
1
14
114
[
22]
P
ol
y
c
hr
on
opo
ul
o
s
D
,
W
e
i
t
s
c
h
ek
E
,
D
i
m
i
t
r
i
ev
a
S
,
B
uc
her
P
,
F
el
i
c
i
G
,
A
l
m
i
r
a
n
t
is
Y
.
C
l
as
s
if
i
c
a
t
i
o
n
of
s
el
e
c
t
i
v
e
l
y
c
on
s
t
r
a
i
ne
d D
N
A
el
em
ent
s
u
s
i
n
g f
e
at
ur
e
v
ec
t
or
s
an
d r
u
l
e
-
ba
s
ed
c
l
as
s
i
f
i
er
s
.
G
enom
i
c
s
.
2014;
104(
2)
:
79
-
86.
[
23]
T
hangar
aj
M
,
V
i
j
ay
al
a
k
s
hm
i
C
R
.
P
er
f
or
m
anc
e S
t
u
dy
on R
ul
e
-
bas
e
d C
l
a
s
s
i
f
i
c
at
i
on T
ec
h
ni
qu
es
a
c
r
os
s
M
ul
t
i
pl
e D
at
ab
as
e R
e
l
at
i
ons
.
I
nt
er
na
t
i
o
nal
J
our
n
al
of
A
p
pl
i
ed I
nf
or
m
at
i
on S
y
s
t
em
s
(
I
J
A
I
S
)
–
I
SSN
.
2013:
224
9
-
086
8.
[
24]
K
han
A
,
B
ah
ar
ud
i
n
B
,
Le
e LH
,
K
han
K
.
A
r
ev
i
ew
of
m
a
c
hi
n
e l
e
ar
ni
ng
al
gor
i
t
h
m
s
f
or
t
ex
t
-
doc
u
m
en
t
s
c
l
as
s
if
i
c
a
t
io
n
.
J
our
n
al
of
ad
v
a
n
c
es
i
n i
nf
or
m
at
i
on t
ec
hnol
og
y
.
2010
;
1
(
1)
:
4
-
20.
[
25]
F
eng X
,
Li
G
,
L
i
na
G
.
A N
e
w
Su
b
-
pi
x
el
E
dge D
e
t
ec
t
i
on
M
et
hod of
C
ol
or
I
m
a
ges
.
I
n
do
n
es
ia
n
J
ou
r
nal
o
f
E
l
e
c
t
r
i
c
al
E
ngi
neer
i
ng and C
om
put
er
S
c
i
en
c
e
.
2014
.
12
(5
)
:
3609
-
361
5.
[
26]
D
haw
an S
.
A
r
ev
i
ew
of
i
m
age
c
om
pr
es
s
i
on a
nd c
om
par
i
s
on
of
i
t
s
al
gor
i
t
hm
s
.
I
nt
er
n
at
i
o
nal
J
our
n
al
o
f
E
l
ec
t
r
oni
c
s
&
C
om
m
uni
c
at
i
on
T
ec
h
nol
o
gy
,
I
J
E
C
T
.
201
1
;
2
(1
):
22
-
6.
Evaluation Warning : The document was created with Spire.PDF for Python.