I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
,
p
p
.
2
4
2
5
~2
4
3
2
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v9
i
4
.
p
p
2
4
2
5
-
2432
2425
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
I
m
pro
v
ing
of
fing
erprint
seg
m
entat
io
n i
m
a
g
es ba
sed
on
K
-
m
ea
ns
a
nd
DB
SCAN
c
lusteri
ng
E
l M
ehdi Cher
ra
t
1
,
Ra
chid Ala
o
ui
2
,
H
a
s
s
a
ne
B
o
uza
hir
3
1,
3
L
a
b
o
ra
to
ry
o
f
S
y
ste
m
s E
n
g
in
e
e
rin
g
a
n
d
I
n
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
,
Na
ti
o
n
a
l
S
c
h
o
o
l
o
f
A
p
p
li
e
d
S
c
ien
c
e
s,
Ib
n
Z
o
h
r
Un
iv
e
rsity
in
A
g
a
d
ir,
M
o
ro
c
c
o
2
L
a
b
o
ra
to
ry
o
f
S
y
ste
m
s
A
n
a
l
y
sis,
In
f
o
rm
a
ti
o
n
P
ro
c
e
ss
in
g
a
n
d
I
n
teg
ra
ted
M
a
n
a
g
e
m
e
n
t
,
Hi
g
h
e
r
S
c
h
o
o
l
o
f
T
e
c
h
n
o
lo
g
y
-
S
a
le
,
M
o
h
a
m
m
e
d
V
U
n
iv
e
rsity
in
Ra
b
a
t,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Au
g
1
,
2
0
1
8
R
ev
i
s
ed
D
ec
1
9
,
2
0
1
8
A
cc
ep
ted
Ma
r
4
,
2
0
1
9
No
w
a
d
a
y
s,
th
e
f
in
g
e
rp
rin
t
id
e
n
ti
f
ica
ti
o
n
sy
ste
m
is
th
e
m
o
st
e
x
p
lo
i
ted
se
c
to
r
o
f
b
io
m
e
tri
c
.
F
in
g
e
rp
ri
n
t
im
a
g
e
se
g
m
e
n
tatio
n
is
c
o
n
sid
e
re
d
o
n
e
o
f
it
s
f
irst
p
ro
c
e
ss
in
g
sta
g
e
.
T
h
u
s,
th
is
sta
g
e
a
ff
e
c
t
s
t
y
p
ica
ll
y
th
e
f
e
a
tu
re
e
x
t
r
a
c
ti
o
n
a
n
d
m
a
tch
in
g
p
ro
c
e
ss
w
h
ich
lea
d
s
t
o
f
in
g
e
rp
rin
t
re
c
o
g
n
i
ti
o
n
sy
ste
m
w
it
h
h
ig
h
a
c
c
u
ra
c
y
.
In
th
is
p
a
p
e
r,
t
h
re
e
m
a
jo
r
ste
p
s
a
re
p
ro
p
o
se
d
.
F
irst,
S
o
b
le
a
n
d
T
o
p
Ha
t
f
il
terin
g
m
e
th
o
d
h
a
v
e
b
e
e
n
u
se
d
to
im
p
ro
v
e
th
e
q
u
a
l
it
y
o
f
th
e
f
in
g
e
rp
rin
t
ima
g
e
s.
T
h
e
n
,
f
o
r
e
a
c
h
lo
c
a
l
b
lo
c
k
in
f
in
g
e
rp
rin
t
i
m
a
g
e
,
a
n
a
c
c
u
ra
te
s
e
p
a
ra
ti
o
n
o
f
th
e
f
o
re
g
ro
u
n
d
a
n
d
b
a
c
k
g
ro
u
n
d
re
g
io
n
i
s
o
b
tain
e
d
b
y
K
-
m
e
a
n
s
c
lu
ste
rin
g
f
o
r
c
o
m
b
in
i
n
g
5
-
d
im
e
n
sio
n
a
l
c
h
a
ra
c
teristics
v
e
c
to
r
(v
a
rian
c
e
,
d
iff
e
re
n
c
e
o
f
m
e
a
n
,
g
ra
d
ien
t
c
o
h
e
re
n
c
e
,
ri
d
g
e
d
irec
ti
o
n
a
n
d
e
n
e
rg
y
sp
e
c
tru
m
).
A
d
d
it
io
n
a
ll
y
,
in
o
u
r
a
p
p
r
o
a
c
h
,
th
e
lo
c
a
l
v
a
rian
c
e
th
re
s
h
o
l
d
in
g
is
u
se
d
to
re
d
u
c
e
c
o
m
p
u
ti
n
g
ti
m
e
f
o
r
se
g
m
e
n
tatio
n
.
F
i
n
a
ll
y
,
w
e
a
re
c
o
m
b
in
e
d
to
o
u
r
sy
ste
m
DBSC
A
N
c
lu
ste
rin
g
w
h
ich
h
a
s
b
e
e
n
p
e
rf
o
rm
e
d
in
o
rd
e
r
t
o
o
v
e
rc
o
m
e
th
e
d
ra
w
b
a
c
k
s
o
f
K
-
m
e
a
n
s
c
las
si
f
ic
a
ti
o
n
in
f
in
g
e
rp
rin
t
im
a
g
e
s
se
g
m
e
n
tatio
n
.
T
h
e
p
ro
p
o
se
d
a
lg
o
rit
h
m
is
tes
ted
o
n
f
o
u
r
d
if
fe
re
n
t
d
a
tab
a
se
s.
Ex
p
e
rime
n
tal
re
su
lt
s
d
e
m
o
n
stra
te
th
a
t
o
u
r
a
p
p
ro
a
c
h
is
sig
n
if
ica
n
tl
y
e
ff
ic
a
c
y
a
g
a
in
st
so
m
e
re
c
e
n
tl
y
p
u
b
li
sh
e
d
tec
h
n
iq
u
e
s
in
term
s
o
f
se
p
a
ra
ti
o
n
b
e
tw
e
e
n
th
e
rid
g
e
a
n
d
n
o
n
-
ri
d
g
e
re
g
io
n
.
K
ey
w
o
r
d
s
:
C
las
s
i
f
icatio
n
DB
SC
A
N
Fin
g
ep
r
in
t
s
eg
m
e
n
tatio
n
i
m
a
g
e
K
-
m
ea
n
s
Ma
ch
i
n
e
lear
n
i
n
g
Co
p
y
rig
h
t
©
2
0
1
9
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
E
l
m
eh
d
i c
h
er
r
at,
L
ab
o
r
ato
r
y
o
f
S
y
s
te
m
s
E
n
g
i
n
e
er
in
g
a
n
d
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
,
Natio
n
al
Sc
h
o
o
l o
f
A
p
p
lied
Scien
ce
s
,
I
b
n
Z
o
h
r
U
n
iv
er
s
it
y
in
Ag
ad
ir
,
Mo
r
o
cc
o
.
E
m
ail: e
l.c
h
er
r
at@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
I
n
p
r
esen
t
w
o
r
ld
,
f
in
g
er
p
r
in
ts
h
av
e
b
ec
o
m
e
a
n
i
m
p
o
r
tan
t
b
io
m
etr
ic
tec
h
n
o
lo
g
y
d
u
e
to
it
s
u
n
iq
u
e
n
es
s
an
d
in
v
ar
ian
ce
to
ev
er
y
p
er
s
o
n
.
Mo
r
eo
v
er
,
w
i
th
t
h
e
p
o
p
u
lar
it
y
o
f
f
in
g
er
p
r
in
ti
n
g
tech
n
o
lo
g
y
,
esp
ec
iall
y
i
n
m
o
b
ile
p
h
o
n
es,
t
h
e
tec
h
n
o
lo
g
y
p
r
ev
io
u
s
l
y
u
s
ed
i
n
th
e
f
ield
o
f
cr
i
m
i
n
al
i
n
v
e
s
ti
g
atio
n
i
s
n
o
w
co
m
m
er
cial
ized
.
T
h
is
b
io
m
etr
ic
tr
ait
i
s
m
o
r
e
u
s
ed
an
d
ac
ce
p
tab
le
b
y
t
h
e
u
s
er
s
b
ec
au
s
e
th
e
ca
p
t
u
r
in
g
d
ev
ic
e
is
r
elativ
el
y
s
m
al
l
an
d
id
en
tific
atio
n
ac
cu
r
ac
y
i
s
co
m
p
ar
ati
v
el
y
v
er
y
h
ig
h
to
o
th
er
b
io
m
etr
ic
r
ec
o
g
n
itio
n
tec
h
n
iq
u
e
s
s
u
c
h
as
th
e
r
etin
a,
ir
is
,
h
a
n
d
g
eo
m
etr
y
,
etc
.
[
1
-
2
]
.
T
h
e
i
m
ag
e
s
eg
m
e
n
tatio
n
is
o
n
e
o
f
t
h
e
m
ai
n
p
r
o
b
le
m
s
i
n
t
h
e
f
ield
o
f
co
m
p
u
ter
v
is
io
n
a
n
d
i
m
a
g
e
p
r
o
ce
s
s
in
g
.
T
h
er
ef
o
r
e,
th
e
f
i
n
g
er
p
r
in
t
s
eg
m
e
n
tatio
n
is
t
y
p
ic
all
y
th
e
f
ir
s
t
an
d
f
o
r
e
m
o
s
t
s
te
p
in
th
e
p
r
o
ce
s
s
o
f
b
io
m
etr
ic
r
ec
o
g
n
itio
n
s
y
s
te
m
b
ased
o
n
f
in
g
er
p
r
in
t.
I
n
ad
d
itio
n
,
th
e
e
f
f
ec
t
o
f
t
h
i
s
s
tep
d
ir
ec
tl
y
af
f
ec
t
s
th
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
s
y
s
te
m
.
T
h
e
g
en
er
al
s
tr
u
ctu
r
e
o
f
f
i
n
g
e
r
p
r
in
t
r
ec
o
g
n
itio
n
s
y
s
te
m
co
n
s
is
ts
o
f
f
o
u
r
m
aj
o
r
s
tep
s
.
I
n
th
e
f
ir
s
t
o
n
e,
th
e
ac
q
u
i
s
itio
n
o
f
f
i
n
g
er
p
r
in
t
i
m
ag
e
is
p
r
o
ce
s
s
o
f
g
etti
n
g
a
d
ig
itali
s
ed
i
m
ag
e
o
f
a
p
er
s
o
n
b
y
u
s
i
n
g
th
e
s
p
ec
i
f
ic
s
en
s
o
r
s
.
T
h
ese
i
m
a
g
es
ca
n
b
e
ac
q
u
ir
ed
in
t
w
o
w
a
y
s
:
o
f
f
lin
e
an
d
liv
e
-
s
ca
n
ac
q
u
is
it
io
n
[
3
-
4
]
.
I
n
th
e
s
ec
o
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t 2
0
1
9
:
2
4
2
5
-
243
2
2426
s
tep
,
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
i
s
allo
w
ed
to
i
m
p
r
o
v
e
o
v
er
all
q
u
alit
y
o
f
th
e
ca
p
t
u
r
ed
i
m
a
g
e.
T
h
r
o
u
g
h
,
it
is
f
r
eq
u
e
n
tl
y
d
if
f
ic
u
lt
to
r
ea
lize
th
is
p
r
o
ce
s
s
b
ec
au
s
e
t
h
e
p
r
esen
ce
o
f
lar
g
e
a
m
o
u
n
t
n
o
i
s
y
ar
ea
s
i
n
th
e
i
m
a
g
e
[
5
-
6
]
.
Af
ter
t
h
at,
th
e
s
e
g
m
en
tatio
n
is
ap
p
lied
.
I
t
is
th
e
p
r
o
ce
s
s
o
f
s
e
p
ar
atio
n
in
i
m
ag
e
i
n
to
t
w
o
r
e
g
io
n
s
:
th
e
r
eg
io
n
o
f
th
e
f
i
n
g
er
p
r
in
t
i
m
ag
e
t
h
a
t
co
n
tai
n
s
a
ll
i
m
p
o
r
tan
t
d
ata
n
ee
d
ed
f
o
r
r
ec
o
g
n
itio
n
is
ca
lled
f
o
r
eg
r
o
u
n
d
r
eg
io
n
,
w
h
ile
t
h
e
r
eg
io
n
s
w
h
ich
h
a
v
e
b
ee
n
th
e
b
lu
r
r
ed
o
r
n
o
is
y
ar
ea
ar
e
ca
lled
b
ac
k
g
r
o
u
n
d
r
eg
io
n
.
I
n
th
e
n
ex
t
s
tep
,
th
e
f
ea
tu
r
es
p
o
in
ts
ar
e
ex
tr
ac
t
ed
f
r
o
m
a
p
r
e
-
p
r
o
ce
s
s
ed
f
in
g
e
r
p
r
in
t
i
m
a
g
e
s
u
ch
as
r
id
g
e
en
d
in
g
a
n
d
b
if
u
r
ca
tio
n
u
n
i
f
o
r
m
l
y
ca
lled
m
in
u
tiae.
I
n
th
e
la
s
t
s
tep
,
g
e
n
er
all
y
,
th
e
m
a
tch
i
n
g
o
f
t
h
e
e
x
tr
ac
ted
t
h
e
f
ea
t
u
r
e
p
o
in
ts
in
o
r
d
er
to
p
er
f
o
r
m
t
h
e
id
en
ti
f
icat
io
n
o
f
th
e
p
er
s
o
n
.
Au
to
m
a
tic
s
e
g
m
en
ta
tio
n
h
as
attr
ac
ted
co
n
s
id
er
ab
le
a
m
o
u
n
t
o
f
r
ea
ch
in
ter
e
s
t
in
th
e
la
s
t
d
ec
ad
e.
T
h
er
ef
o
r
e,
in
t
h
is
p
ap
er
,
i
m
p
r
o
v
ed
f
i
n
g
er
p
r
in
t
s
e
g
m
e
n
tatio
n
m
et
h
o
d
u
s
i
n
g
t
w
o
m
ac
h
i
n
e
l
ea
r
n
in
g
m
o
d
els
i
s
p
r
esen
ted
.
I
n
o
u
r
alg
o
r
ith
m
,
w
e
h
av
e
b
ee
n
u
s
ed
p
ar
ticu
lar
f
ilter
i
n
g
m
et
h
o
d
to
ev
alu
ate
t
h
e
q
u
alit
y
o
f
i
m
a
g
e
a
cq
u
ir
ed
.
A
f
ter
th
at,
th
e
f
i
n
g
er
p
r
in
t
i
m
ag
e
i
s
p
ar
titi
o
n
ed
in
to
n
o
n
-
o
v
er
lap
p
in
g
b
lo
ck
s
o
f
p
ar
ticu
lar
s
ize.
Mo
r
eo
v
er
,
f
o
r
ea
ch
b
lo
ck
,
th
e
f
ea
tu
r
e
v
ec
to
r
is
r
ep
r
esen
ted
b
y
its
:
v
ar
ian
ce
,
d
if
f
er
e
n
ce
o
f
m
ea
n
,
g
r
ad
ien
t
co
h
er
en
ce
,
r
id
g
e
o
r
ien
tatio
n
an
d
en
er
g
y
s
p
e
ctr
u
m
.
Fu
r
t
h
er
m
o
r
e,
t
h
e
lo
ca
l
v
ar
ian
ce
th
r
e
s
h
o
ld
in
g
is
u
s
ed
to
d
is
tin
ct
b
et
w
ee
n
th
e
f
ea
t
u
r
es
w
h
ic
h
w
il
l
b
e
co
m
p
u
ted
o
r
co
n
s
id
er
ed
as
n
u
l
l.
T
h
e
f
ir
s
t
m
ac
h
i
n
e
lear
n
i
n
g
,
K
-
m
ea
n
s
clas
s
i
f
ier
,
is
tr
ai
n
e
d
f
o
r
d
iv
id
in
g
ea
ch
e
x
tr
ac
te
d
f
ea
tu
r
e
i
n
to
t
w
o
cla
s
s
e
s
(
f
o
r
e
g
r
o
u
n
d
ar
ea
an
d
b
ac
k
g
r
o
u
n
d
ar
ea
)
.
Fin
al
l
y
,
t
h
e
s
ec
o
n
d
o
n
e
(
DB
S
C
A
N
c
lu
s
te
r
in
g
)
i
s
u
s
ed
to
r
e
m
o
v
e
s
o
m
e
m
is
c
lass
if
ied
b
lo
ck
s
d
u
e
to
K
-
m
ea
n
s
cla
s
s
i
f
icatio
n
.
T
h
u
s
,
th
e
co
n
to
u
r
s
m
o
o
th
in
g
is
p
er
f
o
r
m
ed
to
en
h
a
n
ce
th
e
i
m
ag
e
s
s
e
g
m
e
n
ted
o
f
f
i
n
g
er
p
r
in
t
s
.
T
h
e
r
est
o
f
t
h
e
p
ap
er
is
s
ep
ar
ated
in
to
f
o
u
r
s
e
ctio
n
s
.
I
n
t
h
e
s
ec
t
io
n
2
,
th
e
r
elate
d
w
o
r
k
s
in
t
h
e
f
ield
ar
e
r
ev
ie
w
ed
.
Sectio
n
3
d
is
cu
s
s
es
th
e
p
r
o
p
o
s
ed
s
eg
m
e
n
tatio
n
al
g
o
r
ith
m
.
E
x
p
er
i
m
en
tal
r
esu
lts
f
o
r
f
o
u
r
d
atab
ases
h
a
v
e
b
ee
n
an
a
l
y
s
ed
an
d
d
is
cu
s
s
ed
i
n
Sectio
n
4
.
Fi
n
all
y
,
t
h
e
co
n
c
lu
s
io
n
i
s
p
r
esen
ted
in
th
e
last
s
ec
tio
n
.
2.
RE
L
AT
E
D
WO
RK
S
T
h
e
f
in
g
er
p
r
in
t
i
m
a
g
e
s
e
g
m
en
tatio
n
is
o
n
e
o
f
t
h
e
p
r
in
cip
al
s
tag
e
f
o
r
au
to
m
ated
f
in
g
er
p
r
in
t
r
ec
o
g
n
itio
n
s
y
s
te
m
.
T
h
is
p
r
e
-
p
r
o
ce
s
s
in
g
s
ta
g
e
allo
w
s
to
s
ep
ar
ate
th
e
f
i
n
g
er
p
r
in
t
r
eg
io
n
f
r
o
m
a
ca
p
tu
r
ed
i
m
ag
e
w
it
h
t
w
o
ar
ea
s
:
f
o
r
eg
r
o
u
n
d
a
n
d
b
ac
k
g
r
o
u
n
d
[
7
]
.
Mo
s
t
ex
i
s
tin
g
tec
h
n
iq
u
es
o
f
s
e
g
m
e
n
tat
i
o
n
ar
e
b
ased
o
n
th
e
f
ea
t
u
r
e
o
f
p
i
x
el
i
n
te
n
s
i
t
y
in
a
b
lo
ck
b
ec
au
s
e
it
i
s
co
m
p
u
tati
o
n
all
y
f
a
s
ter
t
h
an
o
th
er
s
b
ase
d
o
n
l
y
o
n
t
h
e
p
i
x
el
in
te
n
s
it
y
[
8
-
1
0
]
.
Fo
r
f
in
g
er
p
r
i
n
t
s
e
g
m
e
n
tat
io
n
,
th
er
e
ar
e
s
o
v
ar
io
u
s
m
et
h
o
d
s
h
a
v
e
b
ee
n
p
r
o
p
o
s
ed
in
th
e
s
tate
-
of
-
t
h
e
-
ar
t.
Her
e,
w
e
b
r
ief
l
y
r
e
v
ie
w
t
h
ese
m
et
h
o
d
s
.
Li
,
et
al.
[
1
1
]
p
r
o
p
o
s
ed
a
s
eg
m
en
tatio
n
tech
n
iq
u
e
b
y
ca
lcu
l
atin
g
g
r
a
y
co
n
tr
ac
t
a
n
d
Fo
u
r
ie
r
s
p
ec
tr
u
m
en
er
g
y
r
atio
f
o
r
ea
ch
b
lo
ck
in
f
i
n
g
er
p
r
in
t
i
m
ag
e
a
n
d
th
e
n
class
i
f
ied
th
e
s
e
b
lo
ck
b
y
lin
ea
r
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
ap
p
r
o
ac
h
.
Fi
n
all
y
,
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
ar
e
u
s
ed
to
i
m
p
r
o
v
e
th
e
s
eg
m
e
n
ted
i
m
a
g
e.
Ak
r
a
m
,
et
al.
p
r
esen
ted
a
s
eg
m
e
n
tat
io
n
m
et
h
o
d
b
y
co
m
p
u
ti
n
g
m
ea
n
,
v
ar
ian
ce
an
d
g
r
ad
ien
t
d
ev
ia
tio
n
in
f
o
r
m
atio
n
o
f
ea
c
h
b
lo
ck
i
n
f
i
n
g
er
p
r
in
t
i
m
a
g
e.
T
h
e
s
e
g
m
e
n
tatio
n
i
m
a
g
e
o
f
f
in
g
er
p
r
in
t
i
s
o
b
tain
ed
b
y
u
s
i
n
g
th
e
l
in
ea
r
c
lass
if
ier
[
1
2
]
.
L
i
,
e
t
al.
[
1
3
]
s
u
g
g
ested
a
m
eth
o
d
f
o
r
f
in
g
er
p
r
in
t
i
m
a
g
e
s
e
g
m
e
n
t
atio
n
u
s
i
n
g
a
n
o
v
e
l
ap
p
r
o
ac
h
o
f
K
-
Me
an
s
.
t
h
e
f
i
n
g
er
p
r
in
t
i
m
a
g
e
i
s
d
i
v
id
ed
i
n
to
n
o
n
-
o
v
er
lap
p
in
g
b
lo
ck
s
.
F
u
r
th
er
m
o
r
e,
f
o
r
ea
c
h
b
lo
ck
,
th
e
v
ar
ian
ce
,
d
ir
ec
tio
n
an
d
en
er
g
y
s
p
ec
tr
u
m
ar
e
ex
tr
ac
ted
to
co
n
s
tr
u
ct
f
ea
t
u
r
e
v
ec
to
r
s
an
d
th
en
,
class
i
f
ied
t
h
ese
ch
ar
ac
ter
i
s
tics
b
y
K
-
m
ea
n
s
cl
u
s
ter
in
g
al
g
o
r
i
th
m
.
F
i
n
all
y
,
u
s
ed
t
h
e
p
o
s
t
-
p
r
o
ce
s
s
in
g
to
r
e
m
o
v
e
th
e
r
e
m
ai
n
i
n
g
is
o
lated
b
lo
c
k
s
i
n
f
o
r
eg
r
o
u
n
d
o
r
b
ac
k
g
r
o
u
n
d
r
eg
io
n
.
I
n
Ya
n
g
,
et
a
l.
[
1
4
-
1
5
]
h
av
e
b
ee
n
s
u
b
j
ec
ted
a
n
o
v
el
al
g
o
r
ith
m
o
f
f
in
g
er
p
r
in
t
i
m
a
g
es
s
eg
m
e
n
t
atio
n
b
y
u
s
i
n
g
a
n
u
n
s
u
p
er
v
is
ed
lear
n
in
g
m
e
th
o
d
b
ased
o
n
K
-
m
ea
n
s
clas
s
i
f
ier
.
T
h
u
s
,
f
o
r
ea
ch
b
lo
ck
,
t
h
e
a
v
e
r
ag
e
an
d
co
h
er
en
ce
d
ata
i
s
co
m
p
u
ted
i
n
o
r
d
er
to
d
iv
id
e
th
e
i
m
a
g
e
i
n
to
t
w
o
r
e
g
io
n
s
b
y
u
s
in
g
K
-
m
ea
n
s
clu
s
te
r
in
g
ap
p
r
o
ac
h
.
T
h
e
co
r
r
elatio
n
b
ased
f
i
n
g
er
p
r
in
t
i
m
a
g
e
s
e
g
m
e
n
tatio
n
is
u
s
ed
i
n
[
1
6
]
.
Fah
m
y
,
et
al.
[
1
7
]
p
r
o
p
o
s
ed
a
tech
n
iq
u
e
t
h
at
u
ti
li
ze
s
m
o
r
p
h
o
lo
g
ical
p
r
o
ce
s
s
in
g
to
ex
tr
ac
t
t
h
e
f
o
r
eg
r
o
u
n
d
f
r
o
m
th
e
f
i
n
g
er
p
r
in
t
i
m
ag
e.
Af
ter
th
e
d
iv
is
io
n
o
f
i
m
a
g
e
i
n
to
n
o
n
-
o
v
er
lap
p
in
g
b
lo
ck
s
,
t
h
i
s
m
et
h
o
d
u
s
ed
th
e
f
ea
t
u
r
e
v
ec
to
r
f
o
r
ea
ch
b
lo
ck
to
r
ea
lize
f
in
g
er
p
r
in
t
s
eg
m
e
n
tat
io
n
.
T
h
en
,
th
e
ad
ap
tiv
e
t
h
r
esh
o
ld
in
g
i
s
u
s
ed
to
co
n
v
er
t
t
h
e
f
in
g
er
p
r
in
t
i
m
a
g
e
to
a
b
in
ar
y
o
n
e.
Nex
t,
s
o
m
e
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
(
clo
s
in
g
a
n
d
o
p
en
i
n
g
)
ar
e
ap
p
lied
,
to
s
e
g
m
e
n
ted
t
h
e
i
m
ag
e.
Fi
n
all
y
,
t
h
e
co
m
p
le
x
Fo
u
r
ier
s
er
ies
ex
p
a
n
s
io
n
ar
e
p
er
f
o
r
m
ed
to
s
m
o
o
th
th
e
s
e
g
m
en
ted
co
n
to
u
r
.
I
n
t
h
i
s
al
g
o
r
ith
m
,
t
h
e
i
m
a
g
e
i
s
s
ep
ar
ated
in
to
b
lo
ck
s
a
n
d
s
u
b
-
b
lo
ck
s
.
A
f
ter
w
ar
d
s
,
t
h
e
t
h
r
es
h
o
ld
in
g
le
v
el
h
av
e
b
ee
n
ap
p
lied
f
o
r
s
eg
m
en
tatio
n
.
Das
,
et
al.
[
1
8
]
ac
h
ie
v
ed
th
e
f
i
n
g
er
p
r
in
t
s
e
g
m
en
ta
tio
n
b
y
co
m
p
u
ti
n
g
b
lo
ck
b
ased
s
tati
s
ti
cs
an
d
m
o
r
p
h
o
lo
g
ica
l
o
p
er
atio
n
s
.
A
b
b
o
u
d
,
et
al.
[
1
9
]
p
r
esen
ted
a
n
e
w
s
e
g
m
en
t
atio
n
tec
h
n
iq
u
e
b
y
s
tati
s
tical
co
m
p
u
ti
n
g
:
m
ea
n
,
v
ar
ian
ce
a
n
d
co
h
er
e
n
ce
f
ea
tu
r
es
o
f
ea
c
h
b
lo
ck
i
n
f
i
n
g
er
p
r
in
t
i
m
a
g
e
b
ased
o
n
an
a
u
to
m
a
ti
c
th
r
es
h
o
ld
v
al
u
es
an
d
Ot
s
u
’
s
m
et
h
o
d
.
Fin
all
y
,
t
h
e
f
il
in
g
th
e
g
ap
s
is
ap
p
lied
to
r
em
o
v
e
t
h
e
n
o
is
e
i
n
s
o
m
e
r
eg
io
n
s
in
f
o
r
eg
r
o
u
n
d
o
r
b
ac
k
g
r
o
u
n
d
b
y
u
s
i
n
g
p
ar
tic
u
lar
s
ets o
f
r
u
le
s
b
ased
o
n
n
ei
g
h
b
o
r
in
g
r
eg
io
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
I
mp
r
o
vin
g
o
f fin
g
erp
r
in
t seg
men
ta
tio
n
ima
g
es b
a
s
ed
o
n
K
-
mea
n
s
a
n
d
DB
S
C
A
N
…
(
E
l m
eh
d
i Ch
err
a
t
)
2427
3.
P
RO
P
O
SE
D
AP
P
RO
ACH
Ou
r
p
r
o
p
o
s
ed
m
et
h
o
d
is
i
m
p
r
o
v
ed
th
e
f
i
n
g
er
p
r
in
t
i
m
ag
e
s
eg
m
e
n
tatio
n
b
ased
o
n
K
-
Me
an
s
a
n
d
DB
SC
A
N
cl
u
s
ter
i
n
g
.
A
r
o
b
u
s
t
an
d
ef
f
ec
ti
v
e
f
i
n
g
er
p
r
in
t
i
m
a
g
e
s
e
g
m
en
ta
tio
n
al
g
o
r
ith
m
i
s
i
m
p
o
r
ta
n
t
p
h
a
s
e
f
o
r
a
f
in
g
er
p
r
in
t
r
ec
o
g
n
i
tio
n
s
y
s
t
e
m
.
I
n
th
i
s
s
ec
tio
n
,
w
e
d
etai
l
th
e
p
r
o
p
o
s
ed
tech
n
iq
u
e
w
h
i
ch
is
ill
u
s
tr
ated
in
Fig
u
r
e
1
.
T
h
e
d
etails o
f
ea
c
h
p
h
ase
ar
e
r
ep
r
esen
ted
in
t
h
e
f
o
ll
o
w
i
n
g
.
Fig
u
r
e
1
.
B
lo
ck
d
iag
r
a
m
o
f
p
r
o
p
o
s
ed
alg
o
r
ith
m
f
o
r
f
in
g
er
p
r
in
t i
m
ag
e
s
eg
m
e
n
tatio
n
3
.
1
.
P
re
-
pro
ce
s
s
ing
I
n
th
i
s
p
h
ase,
So
b
el
a
n
d
T
o
p
Hat
f
ilter
m
eth
o
d
h
a
v
e
b
ee
n
u
s
ed
to
i
m
p
r
o
v
e
th
e
q
u
ali
t
y
o
f
t
h
e
f
i
n
g
er
p
r
in
t i
m
a
g
e.
So
b
el
s
tr
u
ct
u
r
in
g
o
p
er
ato
r
s
S
o
b
el
x
an
d
S
o
b
el
y
f
o
r
i
m
ag
e
ar
e
r
ep
r
esen
ted
in
(
1
)
.
S
o
b
e
l
x
=
(
-
1
0
1
-
2
0
2
-
1
0
1
)
,
S
o
b
e
l
y
=
(
-
1
-
2
-
1
0
0
0
1
2
1
)
(
1
)
T
h
e
g
r
ad
ien
t G
x
an
d
g
r
ad
ien
t G
y
o
f
p
ix
els ar
e
d
ef
i
n
ed
f
r
o
m
i
m
ag
e
I
mg
b
y
(
2
)
.
G
x
=
S
o
b
e
l
x
*
I
mg
,
G
y
=
S
o
b
e
l
y
*
I
mg
(
2
)
T
h
e
r
esu
lt
o
f
g
r
ad
ien
t
is
co
m
b
in
ed
to
f
i
n
d
t
h
e
ab
s
o
l
u
te
m
a
g
n
i
tu
d
e
(
t
h
e
o
u
tp
u
t
ed
g
e)
.
T
h
is
r
es
u
lt
i
s
d
escr
ib
ed
as f
o
llo
w
s
:
G
(
x
,
y
)
=
√
G
x
2
+
G
y
2
(
3
)
T
h
e
f
in
g
er
p
r
in
t
i
m
a
g
e
i
s
a
m
elio
r
ated
af
ter
n
o
r
m
a
li
s
atio
n
an
d
So
b
el
tech
n
iq
u
e.
Ho
w
e
v
er
,
th
e
f
i
n
g
er
p
r
in
t
i
m
a
g
e
is
m
o
r
e
i
m
p
r
o
v
ed
b
y
u
s
i
n
g
th
e
T
o
p
Hat
tech
n
iq
u
e.
T
h
is
f
ilter
is
a
p
r
o
ce
s
s
th
at
ex
tr
ac
t
s
d
etails
an
d
s
m
all
ele
m
e
n
ts
f
r
o
m
i
m
ag
e.
T
o
p
Hat
f
ilter
i
n
g
i
s
b
ased
o
n
d
ilatio
n
,
er
o
s
io
n
,
o
p
en
in
g
an
d
clo
s
in
g
m
et
h
o
d
.
T
h
e
m
o
r
p
h
o
lo
g
ical
d
ilatio
n
an
d
er
o
s
io
n
o
p
er
atio
n
f
o
r
im
a
g
e
I
mg
o
f
s
ize
x×y
w
it
h
s
tr
u
ct
u
r
i
n
g
ele
m
en
t
Se
ar
e
d
ef
in
ed
b
y
(
4
)
an
d
(
5
)
r
esp
ec
tiv
el
y
:
[
I
mg
⊕
S
e
](
x
,
y
)
=
m
a
x
(
s
,
t
)
∈
S
e
{
I
m
g
(
x
+
s,
y
+
t
)
}
(
4
)
[
I
mg
⊖
S
e
](
x
,
y
)
=
m
i
n
(
s
,
t
)
∈
S
e
{
I
m
g
(
x
+
s,
y
+
t
)
}
(
5
)
T
h
e
o
p
en
in
g
a
n
d
clo
s
in
g
p
r
o
ce
s
s
f
o
r
i
m
ag
e
I
mg
w
it
h
s
t
r
u
ctu
r
i
n
g
e
le
m
e
n
t
S
e
ar
e
d
escr
ib
ed
b
y
co
m
b
i
n
i
n
g
t
h
e
er
o
s
io
n
a
n
d
d
ilatatio
n
o
p
er
atio
n
g
i
v
en
b
y
(
6
)
an
d
(
7
)
r
esp
ec
tiv
el
y
:
I
mg
○
S
e
=
(
I
mg
⊖
)
⊕
(
6
)
I
mg
●
S
e
=
(
I
mg
⊕
)
⊖
(
7
)
T
h
e
o
p
en
in
g
To
p
Ha
top
a
n
d
cl
o
s
in
g
To
p
Ha
t
cl
o
p
er
atio
n
s
f
o
r
i
m
a
g
e
I
m
g
w
i
th
s
tr
u
ct
u
r
in
g
ele
m
e
n
t
Se
ar
e
r
ep
r
esen
ted
by
(
8
)
an
d
(
9
)
r
esp
ec
tiv
el
y
:
T
o
p
H
a
t
op
(I
mg
)
=
I
mg
–
(I
mg
○
)
(
8
)
T
o
p
H
a
t
cl
(I
mg
)
=
I
mg
–
(I
mg
●
)
(
9
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t 2
0
1
9
:
2
4
2
5
-
243
2
2428
3.
2
.
Seg
m
ent
a
t
io
n
Af
ter
t
h
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
p
h
as
e,
th
e
f
in
g
er
p
r
in
t
i
m
a
g
e
is
d
i
v
i
d
ed
in
to
n
o
n
-
o
v
er
lap
p
in
g
lo
ca
l
b
lo
ck
s
o
f
s
ize
w
×
w
.
Fu
r
t
h
er
,
to
ev
er
y
b
lo
ck
,
th
e
c
h
ar
ac
ter
is
tic
v
ec
t
o
r
is
class
if
ied
i
n
to
t
w
o
cla
s
s
es:
f
o
r
eg
r
o
u
n
d
a
n
d
b
ac
k
g
r
o
u
n
d
r
eg
io
n
b
y
u
s
i
n
g
K
-
m
ea
n
s
class
if
icatio
n
.
3.
2
.
1.
Cha
ra
ct
er
is
t
ics ex
t
ra
ct
io
n
T
h
e
ch
ar
ac
ter
is
tic
v
ec
to
r
is
r
e
p
r
esen
ted
,
f
o
r
ea
ch
b
lo
ck
in
f
i
n
g
er
p
r
in
t i
m
a
g
e,
b
y
it
s
th
r
ee
c
ateg
o
r
ies
n
a
m
e
l
y
:
i
m
a
g
e
i
n
ten
s
it
y
b
ased
ch
ar
ac
ter
is
tics
,
g
r
ad
ien
t b
ased
ch
ar
ac
ter
is
tics
a
n
d
r
id
g
e
b
ase
d
ch
ar
ac
ter
is
tics
.
a
.
I
m
a
g
e
inte
ns
it
y
ba
s
ed
cha
ra
ct
er
is
t
ics
T
h
e
ch
an
g
e
i
n
i
n
ten
s
it
y
v
a
lu
e
s
is
u
s
u
all
y
s
p
ec
i
f
ic
alo
n
g
th
e
r
id
g
es
a
n
d
n
o
-
r
id
g
es
w
h
e
n
co
m
p
ar
ed
t
o
b
ac
k
g
r
o
u
n
d
ar
ea
s
i
n
f
i
n
g
er
p
r
in
t
i
m
a
g
e.
Ge
n
er
al
i
m
a
g
e
i
n
te
n
s
it
y
b
a
s
ed
ch
ar
ac
ter
is
tics
ca
n
b
e
u
s
ed
to
d
e
f
i
n
e
th
e
m
o
s
t
in
te
n
s
it
y
s
u
c
h
as
d
if
f
er
en
ce
o
f
m
ea
n
,
w
h
ich
i
s
th
e
d
if
f
er
en
ce
b
et
w
ee
n
th
e
lo
ca
l
in
te
n
s
it
y
m
ea
n
an
d
th
e
g
lo
b
al
in
te
n
s
it
y
m
ea
n
,
a
n
d
v
ar
ian
ce
b
lo
c
k
s
i
n
g
i
v
en
i
m
ag
e
I
mg
o
f
s
ize
x×y
.
T
h
ese
p
r
o
p
r
ieties ar
e
c
o
m
p
u
ted
b
y
(
1
1
)
an
d
(
1
2
)
r
es
p
ec
tiv
el
y
.
(
,
)
=
1
W
2
∑
∑
I
mg
(
x
,
y
)
w
y
=1
w
x
=1
(
1
0
)
ℎ
(
,
)
=
g
M
e
a
n
L
(
,
)
−
g
M
e
a
n
G
(
1
1
)
w
h
er
e
I
mgMeanL
is
t
h
e
m
ea
n
i
n
te
n
s
it
y
o
f
t
h
e
lo
ca
l b
lo
ck
o
f
i
m
a
g
e
an
d
I
mgMeanG
i
s
th
e
m
ea
n
i
n
t
en
s
it
y
o
f
t
h
e
g
lo
b
al
i
m
a
g
e.
ℎ
(
,
)
=
1
2
∑
∑
(
(
,
)
−
)
2
=
1
=
1
(
1
2
)
b.
G
ra
dient
ba
s
e
d c
ha
ra
ct
er
is
t
ics
T
h
e
g
r
ad
ien
t
is
u
ti
lized
to
o
b
tain
th
e
d
ir
ec
tio
n
al
v
ar
iatio
n
in
in
te
n
s
it
y
v
al
u
e
alo
n
g
a
d
i
r
ec
tio
n
o
f
i
m
a
g
e
I
mg
o
f
s
ize
x×y
.
C
h
ar
a
cter
is
tics
s
u
c
h
as
g
r
ad
ien
t
co
h
r
en
ce
a
n
d
r
id
g
e
d
ir
ec
tio
n
ca
n
b
e
class
if
ied
i
n
to
g
r
ad
ien
t
b
ased
ch
ar
ac
ter
i
s
ti
cs.
T
h
e
g
r
ad
ien
t
co
h
er
en
ce
an
d
r
id
g
e
d
ir
ec
tio
n
f
ea
t
u
r
es
ar
e
ca
lcu
lated
by
(
1
6
)
an
d
(
2
0
)
r
esp
ec
tiv
el
y
.
ℎ
ℎ
(
,
)
=
√
(
ℎ
(
,
)
−
ℎ
(
,
)
)
+
4
ℎ
(
,
)
2
ℎ
(
,
)
+
ℎ
(
,
)
(
1
3
)
W
h
er
e
ℎ
(
,
)
=
∑
∑
(
2
(
,
)
)
=
1
=
1
(
1
4
)
ℎ
(
,
)
=
∑
∑
(
2
(
,
)
)
=
1
=
1
(
1
5
)
ℎ
(
,
)
=
∑
∑
(
2
(
,
)
∗
2
(
,
)
)
=
1
=
1
(
1
6
)
T
h
e
g
r
ad
ien
t
G
x
an
d
g
r
ad
ien
t
G
y
ar
e
d
ef
in
ed
b
y
(
3
)
.
1
=
∑
∑
(
2
(
,
)
−
2
(
,
)
)
=
1
=
1
(
1
7
)
2
=
∑
∑
2
G
x
(
x
,
y
)
*
G
y
(
x
,
y
)
=
1
=
1
(
1
8
)
(
,
)
=
1
2
tan
−
1
(
2
1
)
(
1
9
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
I
mp
r
o
vin
g
o
f fin
g
erp
r
in
t seg
men
ta
tio
n
ima
g
es b
a
s
ed
o
n
K
-
mea
n
s
a
n
d
DB
S
C
A
N
…
(
E
l m
eh
d
i Ch
err
a
t
)
2429
ℎ
(
,
)
=
{
π
4
Sq
1
=
0
,
Sq
2
<0
3π
4
Sq
1
=
0
,
Sq
2
≥0
D
(
x
,
y
)
+
π
2
Sq
1
>0
D
(
x
,
y
)
Sq
1
<
0
,
Sq
2
≤0
D
(
x
,
y
)
+π
Sq
1
<
0
,
Sq
2
>0
(
2
0
)
c.
G
ra
dient
ba
s
ed
c
ha
ra
ct
er
is
t
ics
C
h
ar
ac
ter
is
tics
s
u
c
h
as
r
id
g
e
f
r
eq
u
en
c
y
(
en
er
g
y
s
p
ec
tr
u
m
)
w
h
ic
h
is
co
m
p
u
te
s
b
y
u
s
in
g
Fo
u
r
ier
tr
an
s
f
o
r
m
f
o
r
ea
ch
lo
ca
l
b
lo
ck
o
f
i
m
a
g
e
I
mg
o
f
s
ize
x×y
ca
n
b
e
g
r
o
u
p
ed
as
r
id
g
e
b
a
s
ed
ch
ar
ac
ter
is
tic
s
.
T
h
is
f
ea
t
u
r
e
is
co
m
p
u
te
by
(
2
2
)
.
(
,
)
=
∑
∑
(
,
)
−
2
(
+
)
=
1
=
1
(
2
1
)
ℎ
(
,
)
=
√
(
(
,
)
)
2
+
(
(
,
)
)
2
(
2
2
)
w
h
er
e
k
,
l ϵ
{1
,
….
.
w
}
an
d
<
(
x
,
y
)
(
k
,
l)
>=
x
k
+
y
l
.
3
.
2
.
2
.
L
o
ca
l v
a
ri
a
nce
t
hresh
o
ldi
ng
T
h
e
v
ar
ian
ce
lo
ca
l
th
r
esh
o
ld
i
n
g
d
ec
id
es
w
h
et
h
er
o
r
n
o
t
th
e
r
eg
io
n
in
f
i
n
g
er
p
r
in
t
i
m
a
g
e
n
ee
d
s
to
co
m
p
u
te
a
n
o
th
er
c
h
ar
ac
ter
is
ti
c
f
o
r
s
eg
m
e
n
tatio
n
.
T
h
e
lo
ca
l
v
ar
ian
ce
i
s
ca
lc
u
lated
in
g
i
v
e
n
i
m
ag
e
I
mg
o
f
s
ize
x×y
by
(
1
2
)
.
I
f
lo
ca
l
in
te
n
s
it
y
v
ar
ian
ce
i
n
a
w
×
w
i
m
a
g
e
b
lo
ck
is
g
r
ea
ter
t
h
an
0
,
th
e
o
t
h
er
c
h
ar
ac
ter
is
tic
o
f
t
h
e
b
lo
ck
w
ill
b
e
co
m
p
u
ted
o
t
h
er
w
i
s
e
ar
e
co
n
s
id
er
ed
as
n
u
ll.
T
h
er
ef
o
r
e,
f
r
o
m
T
ab
le
1
,
in
DB
1
it
is
o
b
s
er
v
ed
t
h
at
th
e
d
i
f
f
er
en
ce
o
f
t
h
e
p
r
o
ce
s
s
i
n
g
ti
m
e
o
f
o
u
r
p
r
o
p
o
s
ed
s
eg
m
en
tatio
n
is
ac
h
iev
ed
al
m
o
s
t
1
0
s
ec
o
n
d
s
f
o
r
a
n
o
b
tain
ed
f
in
g
er
p
r
in
t
i
m
ag
e.
Ad
d
itio
n
all
y
,
it
is
s
h
o
w
n
th
at
t
h
e
av
er
ag
e
s
e
g
m
e
n
tat
io
n
ti
m
e
i
n
4
d
atab
ases
is
less
th
an
o
th
er
m
e
th
o
d
b
ased
o
n
K
-
m
ea
n
s
s
eg
m
e
n
tatio
n
w
h
ic
h
th
e
b
lo
ck
i
s
r
ep
r
ese
n
ted
j
u
s
t
b
y
3
-
d
i
m
e
n
s
io
n
al
ch
ar
ac
ter
is
tic
v
ec
to
r
[
1
3
]
.
T
ab
le
1
.
C
o
m
p
ar
is
o
n
o
f
s
eg
m
en
tatio
n
t
i
m
e
i
n
s
ec
o
n
d
f
o
r
ea
ch
d
atab
ase
in
FV
C
2
0
0
4
D
a
t
a
b
a
se
s
K
-
m
e
a
n
s w
i
t
h
3
f
e
a
t
u
r
e
s
[
1
3
]
O
u
r
A
l
g
o
r
i
t
h
m w
i
t
h
5
f
e
a
t
u
r
e
s
D
B
1
1
7
,
4
6
7
,
9
5
D
B
2
1
7
,
5
2
1
8
,
6
6
D
B
3
1
7
,
7
8
1
8
,
3
0
D
B
4
1
7
,
6
8
1
9
,
6
6
A
v
g
1
7
,
6
1
1
6
,
1
4
3
.
2
.
3
.
K
-
m
ea
n
s
c
lu
s
t
e
ring
I
n
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
,
K
-
m
ea
n
s
clas
s
if
ier
is
u
tili
ze
d
to
cl
ass
i
f
y
t
h
e
f
i
v
e
e
x
tr
ac
t
ed
c
h
ar
a
cter
is
tics
,
v
ar
ian
ce
,
d
if
f
er
en
ce
o
f
m
ea
n
,
g
r
ad
ien
t
co
h
er
en
ce
,
r
id
g
e
d
ir
ec
tio
n
an
d
en
er
g
y
s
p
ec
tr
u
m
,
f
r
o
m
ea
c
h
lo
ca
l
b
lo
ck
in
f
in
g
er
p
r
in
t
i
m
a
g
e
i
n
o
r
d
er
to
d
is
tin
ct
t
h
e
f
o
r
eg
r
o
u
n
d
ar
ea
to
th
e
n
o
i
s
y
b
ac
k
g
r
o
u
n
d
ar
ea
.
T
h
e
ch
ar
ac
ter
is
tic
ex
tr
ac
tio
n
v
ec
to
r
f
o
r
ea
ch
b
lo
ck
is
r
ep
r
esen
ted
b
y
(
2
3
)
.
On
th
e
o
n
e
h
a
n
d
,
th
e
K
-
m
e
an
s
al
g
o
r
ith
m
is
a
p
o
p
u
lar
ly
u
n
s
u
p
er
v
i
s
ed
m
ac
h
i
n
e
lear
n
in
g
m
o
d
el
s
[
2
0
]
u
s
ed
f
o
r
clu
s
ter
in
g
tec
h
n
iq
u
e
o
f
t
h
e
d
ata
[
2
1
]
.
On
th
e
o
th
er
h
an
d
,
it
i
s
s
i
m
p
li
f
ied
to
im
p
le
m
e
n
tatio
n
,
ea
s
ed
o
f
i
n
ter
p
r
etatio
n
,
f
a
s
ter
an
d
ad
ap
ted
to
s
p
ar
s
e
d
ata.
K
-
m
ea
n
s
cla
s
s
i
f
icatio
n
i
s
r
ep
r
esen
ted
as
f
o
llo
w
s
.
First
l
y
,
th
is
ap
p
r
o
ac
h
s
elec
ts
r
an
d
o
m
l
y
t
h
e
n
u
m
b
er
o
f
clu
s
ter
s
a
n
d
ass
ig
n
s
t
h
e
cl
u
s
ter
w
ith
clo
s
es
t
ce
n
tr
o
id
s
;
t
h
en
it
d
eter
m
i
n
es
ea
ch
d
ata
p
o
in
t
to
t
h
e
n
ea
r
est
ce
n
tr
o
id
s
an
d
f
o
r
ev
er
y
cl
u
s
te
r
,
th
e
n
e
w
ce
n
tr
o
id
s
ar
e
r
ec
alcu
lated
,
an
d
t
h
i
s
p
r
o
ce
s
s
is
r
e
p
ea
ted
u
n
til
s
o
m
e
co
n
d
itio
n
is
v
er
if
ied
[
2
2
]
.
Ch
(
,
)
=
[
ℎ
,
ℎ
,
ℎ
ℎ
,
ℎ
,
ℎ
]
(
2
3
)
3
.
2
.
4
.
DB
SCAN
a
lg
o
rit
h
m
DB
SC
A
N,
Den
s
it
y
-
B
a
s
ed
Sp
atial
C
lu
s
ter
in
g
o
f
A
p
p
licat
io
n
s
w
it
h
No
is
e,
is
an
o
t
h
er
m
o
d
el
o
f
m
ac
h
in
e
lear
n
in
g
an
d
clu
s
ter
i
n
g
a
n
al
y
s
is
al
g
o
r
ith
m
p
r
o
p
o
s
ed
b
y
Ma
r
tin
E
s
ter
et
al.
[
2
3
]
.
DB
SC
A
N
is
o
n
e
o
f
th
e
m
o
s
t
co
m
m
o
n
l
y
u
s
ed
cl
u
s
t
er
an
al
y
s
i
s
al
g
o
r
ith
m
s
.
T
h
is
a
l
g
o
r
ith
m
i
s
d
e
n
s
it
y
-
b
ased
:
g
i
v
e
n
a
s
et
o
f
p
o
i
n
ts
in
a
d
ataset,
th
e
alg
o
r
it
h
m
ca
n
t
h
e
n
ea
r
b
y
p
o
in
ts
ar
e
g
r
o
u
p
ed
to
g
eth
er
(
p
o
in
t
s
w
it
h
m
a
n
y
ad
j
a
ce
n
t p
o
in
ts
)
a
n
d
th
e
o
u
tlier
s
i
n
t
h
e
lo
w
d
en
s
it
y
a
r
ea
ar
e
m
ar
k
ed
.
T
h
e
g
en
er
al
alg
o
r
ith
m
o
f
DB
SC
AN
C
l
u
s
ter
in
g
is
s
h
o
w
n
i
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t 2
0
1
9
:
2
4
2
5
-
243
2
2430
alg
o
r
ith
m
2
.
T
h
e
r
ea
s
o
n
o
f
u
s
i
n
g
DB
SC
A
N
o
v
er
o
th
er
clu
s
te
r
in
g
al
g
o
r
ith
m
s
is
t
h
at
i
t d
o
es n
o
t d
e
m
a
n
d
a
f
i
x
ed
n
u
m
b
er
o
f
g
r
o
u
p
s
i
n
d
atase
t.
I
t a
ls
o
r
ec
o
g
n
izes o
u
tlier
s
a
s
n
o
is
e
w
h
ich
s
i
m
p
l
y
in
tr
o
d
u
ce
s
t
h
e
m
i
n
to
t
h
e
clu
s
ter
ev
en
if
th
e
d
ata
p
o
i
n
ts
ar
e
v
er
y
d
if
f
er
e
n
t.
I
n
ad
d
itio
n
,
i
t
is
a
g
o
o
d
p
lace
to
f
in
d
g
r
o
u
p
s
o
f
a
n
y
s
ize
a
n
d
s
h
ap
e.
I
n
o
u
r
p
r
o
p
s
ed
alg
o
r
ith
m
,
D
B
SC
A
N
C
l
u
s
ter
i
n
g
is
u
s
ed
t
o
ac
h
iev
e
m
o
r
e
co
m
p
ac
t
b
lo
ck
s
f
o
r
r
ed
u
ci
n
g
th
e
m
is
c
lass
if
ica
tio
n
r
eg
io
n
d
u
e
to
th
e
K
-
m
ea
n
s
al
g
o
r
ith
m
.
F
ig
u
r
e
2
p
r
esen
t
s
th
e
r
e
s
u
l
t
u
s
in
g
t
h
e
DB
SC
AN
alg
o
r
ith
m
.
Fin
a
l
y
,
th
e
co
n
to
u
r
s
m
o
o
t
h
in
g
(
f
ilter
i
n
g
i
n
a
co
m
p
lex
Fo
u
r
ier
tr
an
s
f
o
r
m
d
o
m
ai
n
[
2
4
]
)
as
p
o
s
t
-
p
r
o
ce
s
s
in
g
tec
h
n
iq
u
e
to
s
m
o
o
th
en
t
h
e
ed
g
e
s
o
f
m
a
s
k
[
1
7
]
.
Alg
o
r
ith
m
1
DB
S
CA
N Clu
str
in
g
I
NP
UT
:
Da
ta
s
et
,
ep
s
,
m
in
_
p
o
in
ts
w
here
Da
ta
s
et
=
s
et
o
f
cla
s
s
i
f
ied
in
s
tan
ce
s
,
ep
s
=d
is
ta
n
ce
,
m
i
n
_
p
o
in
ts
=
m
in
i
m
u
m
n
u
m
b
er
o
f
p
o
in
ts
to
cr
ea
te
d
en
s
e
r
eg
io
n
.
O
UT
P
UT
:
O
u
tp
u
t a
l
l c
lu
s
ter
s
in
Data
s
et
m
ar
k
ed
w
it
h
C
lu
s
t
er
_
L
ab
el
o
r
n
o
is
e
pro
ce
du
re
m
ar
k
all
clu
s
ter
in
Data
s
et
as u
n
v
is
ited
C
lu
s
ter
_
L
ab
el
←
1
f
o
r
ea
ch
u
n
v
i
s
ited
clu
s
ter
x
in
Data
s
et
do
Z
Fi
n
d
Neig
h
b
o
u
r
s
(
x
,
ep
s
,
m
i
n
_
p
o
in
ts
)
if
|
Z
|
<
m
in
_
p
o
in
t
s
t
hen
m
ar
k
x
as
n
o
is
e
else
m
ar
k
x
an
d
ea
ch
cl
u
s
ter
o
f
Z
w
it
h
C
l
u
s
ter
_
L
ab
el
q
u
e
u
eLis
t
←
all
u
n
v
is
ited
clu
s
ter
s
o
f
Z
un
t
il
q
u
eu
eLis
t
is
e
m
p
t
y
do
y
←
d
elete
a
clu
s
ter
f
r
o
m
q
u
e
u
e
L
is
t
Z
←
Fin
d
Nei
g
h
b
o
u
r
s
(
y
,
ep
s
,
m
in
_
p
o
in
ts
)
If
|
Z
|
m
i
n
p
ts
t
he
n
f
o
r
ea
ch
clu
s
ter
w
in
Z
do
m
ar
k
w
w
it
h
C
l
u
s
ter
_
L
ab
el
if
w
is
u
n
v
is
ited
q
u
eu
eLis
t
←
w
q
u
eu
eLis
t
end f
o
r
m
ar
k
y
as v
is
ited
end until
end if
m
ar
k
x
as v
is
i
ted
C
lu
s
ter
_
L
ab
el
←
C
l
u
s
t
er
_
L
ab
el+
1
end f
o
r
end pro
ce
du
re
Fig
u
r
e
2
.
R
e
m
o
v
in
g
t
h
e
m
i
s
cl
ass
i
f
icatio
n
r
eg
io
n
b
y
DB
S
C
AN
alg
o
r
ith
m
:
(
a)
Or
ig
in
al
i
m
a
g
e;
(
b
)
Ou
r
K
-
m
ea
n
s
clas
s
i
f
icatio
n
; (
c)
Ou
r
K
-
m
ea
n
s
an
d
DB
S
C
A
N
clas
s
i
f
icatio
n
4.
RE
SU
L
T
S
A
ND
D
I
SCU
SS
I
O
N
T
h
e
ex
p
er
im
e
n
tal
o
p
er
atio
n
p
latf
o
r
m
i
n
th
is
s
t
u
d
y
i
s
d
escr
ib
ed
as
f
o
llo
w
s
:
t
h
e
h
o
s
t
co
n
f
ig
u
r
at
io
n
:
C
P
U
I
n
tel
C
o
r
e2
D
u
o
at
2
.
0
0
GHz
,
R
AM
3
.
0
0
GB
,
r
u
n
ti
m
e
en
v
ir
o
n
m
e
n
t:
Mic
r
o
s
o
f
t
Vi
s
u
a
l
St
u
d
io
C
++
2
0
1
3
w
it
h
Op
en
C
V
l
ib
r
ar
y
.
I
n
o
r
d
er
to
b
etter
v
er
if
y
o
u
r
al
g
o
r
ith
m
,
t
h
e
f
o
llo
w
in
g
s
e
g
m
e
n
tatio
n
m
et
h
o
d
s
ar
e
ad
o
p
ted
in
th
e
ex
p
er
i
m
e
n
t:
S
VM
[
9
]
,
K
-
m
ea
n
s
w
it
h
3
-
d
i
m
en
s
io
n
f
ea
tu
r
e
[
1
3
]
,
M
P
[
1
7
]
,
A
C
T
[
1
9
]
.
T
h
ese
s
eg
m
e
n
tatio
n
al
g
o
r
i
th
m
s
w
er
e
co
m
p
ar
ed
to
ea
ch
o
th
er
.
I
n
o
r
d
er
to
v
alid
ate
th
e
p
r
o
p
o
s
ed
al
g
o
r
ith
m
,
th
e
r
esu
lts
h
av
e
b
ee
n
tes
ted
o
n
th
e
p
u
b
lic
Fin
g
er
p
r
in
t
Ver
i
f
icatio
n
C
o
m
p
etitio
n
2
0
0
4
d
ataset
[
2
5
]
w
h
ic
h
co
n
ta
in
s
4
d
atab
ases
,
n
a
m
el
y
DB
l,
DB
2
,
DB
3
an
d
DB
4
.
T
h
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
is
u
s
e
d
th
e
n
u
m
b
er
o
f
m
is
c
lass
if
ica
tio
n
as d
e
f
in
ed
b
y
(
2
5
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
I
mp
r
o
vin
g
o
f fin
g
erp
r
in
t seg
men
ta
tio
n
ima
g
es b
a
s
ed
o
n
K
-
mea
n
s
a
n
d
DB
S
C
A
N
…
(
E
l m
eh
d
i Ch
err
a
t
)
2431
1
=
,
2
=
(
2
4
)
Pr
ob
E
r
r
=
A
vg(
1
,
12
)
(
2
5
)
w
h
er
e
N
b
r
be
is
n
u
m
b
er
o
f
b
ac
k
g
r
o
u
n
d
cla
s
s
i
f
ied
er
r
o
r
,
Nb
rb
is
t
h
e
to
tal
n
u
m
b
er
o
f
tr
u
e
b
ac
k
g
r
o
u
n
d
r
e
g
io
n
s
i
n
th
e
fi
n
g
er
p
r
in
t
i
m
ag
e
a
n
d
P
r
o
b
1
is
th
e
p
r
o
b
ab
ilit
y
th
a
t
a
f
o
r
eg
r
o
u
n
d
r
eg
io
n
is
clas
s
i
f
ied
as
b
ac
k
g
r
o
u
n
d
.
N
b
r
fe
is
th
e
n
u
m
b
er
o
f
f
o
r
eg
r
o
u
n
d
class
i
f
ied
er
r
o
r
,
N
b
r
f
is
th
e
to
tal
n
u
m
b
er
o
f
tr
u
e
f
o
r
eg
r
o
u
n
d
r
eg
io
n
s
i
n
t
h
e
f
i
n
g
er
p
r
in
t
i
m
ag
e
a
n
d
P
r
o
b
2
is
t
h
e
p
r
o
b
ab
ilit
y
t
h
at
a
b
ac
k
g
r
o
u
n
d
r
e
g
io
n
is
cla
s
s
i
f
ied
a
s
f
o
r
eg
r
o
u
n
d
.
T
h
e
p
r
o
b
a
b
ilit
y
o
f
er
r
o
r
P
r
o
b
Err
is
th
e
av
er
a
g
e
o
f
P
r
o
b
1
an
d
P
r
o
b
2
.
T
h
e
ex
is
t
in
g
a
lg
o
r
it
h
m
s
o
f
f
i
n
g
er
p
r
in
t
i
m
a
g
e
s
eg
m
e
n
tatio
n
,
SVM
[
1
1
]
,
K
-
Me
a
n
s
w
it
h
3
-
d
i
m
e
n
s
io
n
f
ea
t
u
r
e
[
1
3
]
,
MP
[
1
7
]
an
d
A
T
C
[
1
9
]
,
ar
e
im
p
le
m
e
n
ted
f
o
r
co
m
p
ar
is
o
n
o
f
s
e
g
m
en
t
atio
n
p
er
f
o
r
m
a
n
ce
.
T
h
e
co
m
p
ar
is
o
n
b
e
t
w
ee
n
p
r
o
p
o
s
ed
m
e
th
o
d
a
n
d
o
th
er
s
w
o
r
k
s
i
n
m
ea
s
u
r
e
o
f
a
v
er
ag
e
s
eg
m
en
tatio
n
er
r
o
r
f
o
r
f
i
n
g
er
p
r
in
t
i
m
ag
e
s
at
d
i
f
f
er
en
t
d
atab
ases
is
s
h
o
w
n
in
T
ab
le
2
.
Fro
m
T
ab
le
2
,
in
T
h
e
SVM
[
1
1
]
an
d
K
-
Me
an
s
w
it
h
3
-
d
i
m
e
n
s
io
n
f
ea
t
u
r
e
[
1
3
]
,
th
e
co
r
r
esp
o
n
d
in
g
v
alu
e
f
o
r
m
is
c
lass
if
ica
tio
n
r
ate
i
n
DB
1
is
1
8
,
7
5
%,
2
0
,
2
8
%
r
esp
ec
tiv
el
y
.
F
u
r
th
er
m
o
r
e,
th
e
er
r
o
r
r
ate
o
f
s
e
g
m
en
ta
tio
n
in
MP
[
1
7
]
,
A
T
C
[
1
9
]
an
d
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
is
0
,
2
8
%,1
3
,
3
1
%
an
d
0
,
3
0
%
r
e
s
p
ec
tiv
el
y
.
I
n
t
h
e
s
ec
o
n
d
d
at
ab
ase,
if
SVM
[
1
1
]
,
K
-
Me
a
n
s
w
it
h
3
-
d
i
m
e
n
s
io
n
f
ea
t
u
r
e
[
1
3
]
,
M
P
[
1
7
]
an
d
A
T
C
[
1
9
]
g
iv
e
3
4
,
5
6
%,
2
2
,
3
0
%,
2
,
9
2
%,
2
9
,
7
9
%
r
esp
ec
tiv
el
y
in
s
eg
m
e
n
tatio
n
r
ate
h
o
w
ev
er
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
tak
es
m
i
n
i
m
u
m
er
r
o
r
r
ate
as
1
,
6
6
%.
L
ik
e
w
is
e,
i
n
th
e
la
s
t
d
ata
b
ase,
m
is
c
lass
if
ica
tio
n
r
ate
(
1
,
0
9
%)
is
clea
r
ly
les
s
f
o
r
th
e
p
r
o
p
o
s
e
d
alg
o
r
ith
m
co
m
p
ar
ed
to
o
th
er
tech
n
iq
u
e
s
in
th
e
th
ir
d
d
atab
ase.
SVM
[
1
1
]
h
a
s
f
a
iled
to
clas
s
if
y
b
etter
t
h
e
f
o
r
w
o
u
n
d
an
d
b
ac
k
g
r
o
u
n
d
r
eg
io
n
(
er
r
o
r
r
ate
as
2
8
,
8
9
%)
in
th
e
f
o
u
r
th
d
atab
ase.
th
u
s
,
o
u
r
s
y
s
te
m
h
a
s
s
u
cc
e
ed
ed
to
r
ed
u
ce
0
,
6
3
%
o
f
s
e
g
m
en
tatio
n
er
r
o
r
th
a
n
K
-
Me
a
n
s
w
it
h
3
-
d
i
m
e
n
s
io
n
s
f
ea
t
u
r
es
[
1
3
]
,
MP
[
1
7
]
an
d
AT
C
[
1
9
]
w
h
ic
h
g
i
v
e
5
,
0
6
%,
1
,
3
1
%
an
d
1
7
,
3
6
%
r
esp
ec
tiv
el
y
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
r
e
s
u
lts
s
h
o
w
a
b
ette
r
p
er
f
o
r
m
an
ce
r
esp
ec
t
to
o
th
er
alg
o
r
ith
m
s
w
i
th
av
er
ag
e
m
is
cla
s
s
i
f
icat
io
n
r
ate
as
0
,
6
7
%
at
d
if
f
er
en
t
d
atab
ases
o
f
f
in
g
er
p
r
in
t
i
m
a
g
es.
T
h
e
m
ea
n
s
o
f
th
e
s
e
r
esu
lt
s
s
h
o
w
s
th
a
t
o
u
r
alg
o
r
it
h
m
i
s
i
m
p
r
o
v
ed
t
h
e
r
ec
o
g
n
i
ti
o
n
ac
cu
r
ate
r
ate
o
f
th
e
p
er
s
o
n
.
Fro
m
t
h
ese
e
x
i
s
ti
n
g
w
o
r
k
s
,
it
is
w
o
r
th
y
to
s
a
y
th
at
th
e
r
esu
l
ts
o
f
p
r
o
p
o
s
ed
m
et
h
o
d
ar
e
clea
r
ly
s
u
p
er
io
r
in
te
r
m
s
eg
m
e
n
tatio
n
er
r
o
r
r
ate.
T
h
e
F
ig
u
r
e
3
r
ep
r
esen
ts
th
e
v
i
s
u
al
r
es
u
lt
s
o
f
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
a
n
d
th
e
o
th
er
tec
h
n
iq
u
es.
W
h
e
n
w
e
co
m
p
ar
e
d
to
t
h
e
e
x
i
s
ti
n
g
w
o
r
k
s
i
n
t
h
is
f
ig
u
r
e
.
W
e
ca
n
s
a
y
th
at
o
u
r
p
r
o
p
o
s
ed
is
e
f
f
ic
ien
t
an
d
r
ed
u
ce
d
er
r
o
r
r
ate
of
s
eg
m
e
n
tatio
n
.
T
h
e
v
is
u
al
q
u
alit
y
r
esu
lts
o
f
s
eg
m
e
n
ted
im
ag
e
i
n
d
icate
t
h
at
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
ad
ap
t
s
an
d
g
i
v
es
b
etter
r
esu
l
ts
i
n
ter
m
o
f
s
e
g
m
e
n
tat
io
n
i
n
d
i
f
f
er
e
n
t
e
n
v
ir
o
n
m
e
n
t
s
t
h
an
th
e
co
m
p
ar
at
iv
e
tech
n
iq
u
e
s
f
o
r
ac
cu
r
ac
y
s
e
g
m
en
tatio
n
.
T
ab
le
2
.
C
o
m
p
ar
is
o
n
o
f
s
eg
m
en
tatio
n
m
i
s
clas
s
i
f
icatio
n
u
s
i
n
g
d
if
f
er
en
t a
l
g
o
r
ith
m
s
in
FVC
2
0
0
4
D
a
t
a
b
a
se
s
S
V
M
[
1
1
]
K
-
m
e
a
n
s [1
3
]
M
P
[
1
7
]
A
T
C
[
1
9
]
P
r
o
p
o
se
d
A
l
g
o
r
i
t
h
m
D1
D2
D3
D4
A
v
g
1
8
,
7
5
%
3
4
,
5
6
%
1
,
3
2
%
2
8
,
8
9
%
2
8
,
3
8
%
2
0
,
2
8
%
22
,
3
0
%
1
2
,
6
0
%
5
,
0
6
%
1
5
,
0
6
%
0
,
2
8
%
2
,
9
2
%
2
,
5
4
%
1
,
3
1
%
1
,
7
6
%
1
3
,
3
1
%
2
9
,
7
9
%
1
9
.
4
4
%
1
7
,
3
6
%
1
9
,
9
8
%
0
,
3
0
%
1
,
6
6
%
1
,
0
9
%
0
,
6
3
%
0
,
9
2
%
Fig
u
r
e
3
.
Seg
m
en
ta
tio
n
r
es
u
lts
u
s
i
n
g
d
i
f
f
er
en
t
m
et
h
o
d
s
in
D
B
2
:
(
a)
Or
ig
in
al
i
m
a
g
e,
(
b
)
SVM
[
1
1
]
,
(
c
)
K
-
m
ea
n
s
[
1
3
]
,
(
d
)
MP
[
1
7
]
,
(
e
)
A
T
C
[
1
9
]
,
(
f
)
P
r
o
p
o
s
ed
m
et
h
o
d
5.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
h
as p
r
o
p
o
s
ed
an
i
m
p
r
o
v
ed
m
e
th
o
d
o
f
f
i
n
g
er
p
r
in
t
s
eg
m
e
n
tatio
n
i
m
a
g
e
s
b
ased
o
n
K
-
Me
a
n
s
class
i
f
icatio
n
an
d
DB
S
C
A
N
clu
s
ter
i
n
g
.
Ou
r
p
r
o
p
o
s
ed
s
y
s
t
e
m
is
p
r
ese
n
ted
in
t
h
r
ee
s
tep
s
.
I
n
th
e
f
ir
s
t
s
tep
,
th
e
q
u
alit
y
o
f
f
in
g
er
p
r
in
t
i
m
a
g
es
is
e
n
h
an
ce
d
u
s
in
g
So
b
le
an
d
T
o
p
Hat
f
ilter
in
g
m
et
h
o
d
.
I
n
t
h
e
s
ec
o
n
d
s
tep
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t 2
0
1
9
:
2
4
2
5
-
243
2
2432
K
-
m
ea
n
s
ap
p
r
o
ac
h
is
ap
p
lied
f
o
r
ea
ch
lo
ca
l
b
lo
ck
to
class
if
y
th
e
i
m
a
g
e
in
to
f
o
r
eg
r
o
u
n
d
an
d
b
ac
k
g
r
o
u
n
d
r
eg
io
n
u
s
i
n
g
f
i
v
e
-
d
i
m
e
n
s
io
n
al
c
h
ar
ac
ter
is
tic
v
ec
to
r
e
x
tr
a
ctio
n
.
Mo
r
eo
v
er
,
t
h
e
p
r
o
ce
s
s
in
g
ti
m
e
o
f
s
eg
m
e
n
tatio
n
is
f
as
ter
th
a
n
a
n
o
th
er
alg
o
r
it
h
m
b
ased
o
n
j
u
s
t
3
d
i
m
en
s
io
n
s
f
ea
tu
r
es
d
u
e
to
t
h
e
lo
ca
l
v
ar
ian
ce
th
r
es
h
o
ld
in
g
.
I
n
th
e
f
i
n
al
s
tep
,
DB
SC
A
N
alg
o
r
it
h
m
i
s
u
s
ed
to
r
em
o
v
e
s
o
m
e
m
is
cla
s
s
i
f
ie
d
b
lo
ck
s
d
u
e
to
th
e
K
-
m
ea
n
s
clu
s
ter
in
g
an
d
co
n
to
u
r
s
m
o
o
th
in
g
is
ac
h
ie
v
ed
to
im
p
r
o
v
e
th
e
i
m
a
g
e
s
e
g
m
en
ted
.
Si
m
u
latio
n
r
es
u
lt
s
s
h
o
w
s
ig
n
i
f
ica
n
tl
y
t
h
at
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
is
ef
f
icac
y
s
o
m
e
r
ec
en
t
ex
is
ti
n
g
tec
h
n
iq
u
e
s
i
n
av
er
a
g
e
s
eg
m
e
n
tatio
n
er
r
o
r
r
ate.
T
h
er
ef
o
r
e,
it
af
f
ec
ts
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
s
y
s
te
m
to
h
av
e
a
h
i
g
h
er
ac
c
u
r
ate
r
ec
o
g
n
itio
n
r
ate
o
f
th
e
p
er
s
o
n
.
RE
F
E
R
E
NC
E
S
[1
]
E.
M
C
h
e
rra
t,
R.
A
lao
u
i
,
H.
B
o
u
z
a
h
ir,
W
.
Je
n
k
a
l
,
"
Hig
h
d
e
n
sity
sa
lt
-
a
n
d
-
p
e
p
p
e
r
n
o
ise
su
p
p
re
ss
io
n
u
sin
g
a
d
a
p
ti
v
e
d
u
a
l
th
re
sh
o
l
d
d
e
c
isio
n
b
a
se
d
a
lg
o
rit
h
m
in
f
in
g
e
rp
ri
n
t
im
a
g
e
s
,
"
IEE
E
In
telli
g
e
n
t
S
y
ste
ms
&
Co
mp
u
ter
Vi
sio
n
,
p
p
.
1
-
4
,
2
0
1
7
.
[2
]
S
.
R.
Bo
rra
,
G
.
J
Re
d
d
y
,
E.
S
.
Re
d
d
y
,
"
A
n
E
ff
icie
n
t
F
in
g
e
rp
rin
t
Id
e
n
ti
f
ica
ti
o
n
u
si
n
g
Ne
u
ra
l
Ne
tw
o
rk
a
n
d
BAT
A
l
g
o
rit
h
m
,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
E
n
g
in
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
8
,
n
o
.
2
,
p
p
.
1
1
9
4
-
1
2
1
3
,
2
0
1
8
.
[3
]
P
.
W
a
n
g
,
"
P
a
tt
e
rn
re
c
o
g
n
it
io
n
,
m
a
c
h
in
e
in
telli
g
e
n
c
e
a
n
d
b
io
m
e
tri
c
s
,
"
S
p
rin
g
e
r
Be
rli
n
He
i
d
e
lb
e
rg
,
2
0
1
1
.
[4
]
A
.
Ro
ss
a
n
d
J.
A
n
il
,
"
Bio
m
e
tri
c
s
e
n
so
r
in
tero
p
e
ra
b
il
it
y
:
A
c
a
se
stu
d
y
in
f
in
g
e
rp
rin
ts
,
"
In
ECCV
W
o
rk
sh
o
p
Bi
o
AW
,
p
p
.
1
3
4
-
145
,
2
0
0
4
.
[5
]
M
.
N.
A
b
d
u
lw
a
h
e
d
a
n
d
A
.
k
a
m
il
A
h
m
e
d
,
"
U
n
d
e
rwa
ter
I
m
a
g
e
d
e
-
n
o
sin
g
u
si
n
g
d
isc
re
te
w
a
v
e
let
tra
n
sf
o
rm
a
n
d
p
re
-
w
h
it
e
n
in
g
f
il
ter
,
"
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
ica
ti
o
n
Co
mp
u
ti
n
g
E
lec
tro
n
ics
a
n
d
C
o
n
tr
o
l
,
v
o
l
.
1
6
,
n
o
.
6
,
2
0
1
8
.
[6
]
R.
M
.
Zay
e
d
,
"
Op
ti
m
u
m
I
m
a
g
e
F
i
l
ters
f
o
r
V
a
rio
u
s
T
y
p
e
s
o
f
No
ise
,
"
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
ica
ti
o
n
C
o
mp
u
ti
n
g
El
e
c
tro
n
ics
a
n
d
C
o
n
tro
l
,
v
o
l.
1
6
,
n
o
.
5
,
2
0
1
8
.
[7
]
B.
S
t
o
jan
o
v
i,
O.
M
a
rq
u
e
s,
a
n
d
A.
Ne
k
o
v
i
,
"
L
a
ten
t
o
v
e
rlap
p
e
d
f
in
g
e
rp
rin
t
se
p
a
ra
ti
o
n
:
a
re
v
iew
,
"
M
u
lt
ime
d
i
a
T
o
o
ls
a
n
d
Ap
p
li
c
a
ti
o
n
s
,
7
6
(
1
5
)
,
p
p
.
1
6
2
6
3
-
1
6
2
9
0
,
2
0
1
7
.
[8
]
C.
W
u
,
S
.
T
u
ly
a
k
o
v
,
V
.
G
o
v
in
d
a
ra
ju
,
"
Ro
b
u
st
p
o
in
t
-
b
a
se
d
f
e
a
tu
re
fin
g
e
rp
rin
t
se
g
m
e
n
tatio
n
a
lg
o
rit
h
m
,
"
Ad
v
a
n
c
e
s
in
Bi
o
me
trics
,
p
p
.
1
0
9
5
-
1
1
0
3
,
2
0
0
7
.
[9
]
E.
Zh
u
,
J.
Yin
,
C.
Hu
,
G
.
Z
h
a
n
g
,
"
A
s
y
ste
m
a
ti
c
m
e
th
o
d
f
o
r
f
in
g
e
rp
rin
t
ri
d
g
e
o
rien
tati
o
n
e
stim
a
ti
o
n
a
n
d
im
a
g
e
se
g
m
e
n
tatio
n
,
"
.
P
a
tt
e
rn
Rec
o
g
n
it
i
o
n
,
3
9
(8
),
p
p
.
1
4
5
2
-
1
4
7
2
,
2
0
0
6
.
[1
0
]
N.
V
.
B
o
u
lg
o
u
ris,
K
.
N.
P
lata
n
io
t
is
,
E.
M
ich
e
li
-
T
z
a
n
a
k
o
u
,
"
Bio
m
e
tri
c
s:
th
e
o
ry
,
m
e
th
o
d
s,
a
n
d
a
p
p
li
c
a
ti
o
n
s
,
"
J
o
h
n
W
il
e
y
&
S
o
n
s
,
(
Vo
l.
9
)
,
2
0
0
9
.
[1
1
]
Y.
X
.
L
i,
X
.
Q.
P
e
n
g
a
n
d
S
.
L
i,
"
F
in
g
e
rp
rin
t
im
a
g
e
s
e
g
m
e
n
tatio
n
m
e
th
o
d
b
a
se
d
o
n
li
n
e
a
r
su
p
p
o
r
t
v
e
c
to
r
m
a
c
h
in
e
,
"
Ap
p
li
c
a
ti
o
n
Res
e
a
rc
h
o
f
C
o
mp
u
te
rs
,
2
5
(
1
2
),
p
p
.
3
6
6
9
-
3
6
7
0
,
2
0
0
8
.
[1
2
]
M
.
U.
A
k
ra
m
,
A
.
Ay
a
z
a
n
d
J.
Im
ti
a
z
,
"
M
o
rp
h
o
l
o
g
ica
l
a
n
d
g
ra
d
ien
t
b
a
se
d
f
in
g
e
rp
rin
t
im
a
g
e
se
g
m
e
n
ta
ti
o
n
"
,
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
I
n
fo
r
ma
ti
o
n
a
n
d
Co
mm
u
n
ica
ti
o
n
T
e
c
h
n
o
l
o
g
ies
(
ICICT
)
,
p
p
.
1
4
,
2
0
1
1
.
[1
3
]
H.
L
i
a
n
d
Y.
P
in
g
,
"
A
n
o
v
e
l
a
p
p
r
o
a
c
h
o
f
K
-
m
e
a
n
s
b
a
se
d
f
in
g
e
rp
rin
t
se
g
m
e
n
tatio
n
a
lg
o
rit
h
m
"
,
In
In
stru
me
n
ta
ti
o
n
&
M
e
a
su
re
me
n
t,
S
e
n
so
r Ne
tw
o
rk
a
n
d
Au
to
ma
ti
o
n
,
IE
EE
I
n
ter
n
a
t
io
n
a
l
S
y
mp
o
si
u
m o
n
,
Vo
l.
1
,
p
p
.
2
1
8
-
2
2
1
,
2
0
1
2
.
[1
4
]
G
.
Y
a
n
g
,
G.
-
T
.
Zh
o
u
,
Y.
Yi
n
,
a
n
d
X
.
Ya
n
g
,
"
K
-
M
e
a
n
s
Ba
se
d
F
in
g
e
rp
rin
t
S
e
g
m
e
n
tatio
n
w
it
h
S
e
n
so
r
In
tero
p
e
ra
b
il
it
y
,
"
EURA
S
IP
J
.
A
d
v
.
S
ig
n
a
l
Pro
c
e
ss
,
v
o
l.
2
0
1
0
,
n
o
.
1
,
p
.
7
2
9
3
7
8
,
2
0
1
0
.
[1
5
]
G
.
Ya
n
g
,
Y.
L
i,
Y.
Yin
,
a
n
d
Y.
-
S
.
L
i,
"
Tw
o
-
L
e
v
e
l
Ev
a
lu
a
ti
o
n
o
n
S
e
n
so
r
In
ter
o
p
e
ra
b
il
it
y
o
f
F
e
a
tu
re
s
in
F
in
g
e
rp
ri
n
t
Im
a
g
e
S
e
g
m
e
n
tatio
n
,
"
S
e
n
so
rs
,
v
o
l.
1
2
,
n
o
.
1
2
,
p
p
.
3
1
8
6
3
1
9
9
,
2
0
1
2
.
[1
6
]
S
.
R.
S
u
ra
lk
a
r
a
n
d
P
.
M
.
P
a
ti
l
,
"
Co
rre
latio
n
b
a
se
d
F
in
g
e
rp
ri
n
t
Im
a
g
e
S
e
g
m
e
n
tatio
n
,
"
In
t.
J
.
Co
m
p
u
t
.
Ap
p
l
.
,
v
o
l
.
6
8
,
n
o
.
7
,
p
p
.
1
3
,
2
0
1
3
.
[1
7
]
M
.
F
.
F
a
h
m
y
a
n
d
M
.
A
.
T
h
a
b
e
t,
"
A
f
in
g
e
rp
rin
t
se
g
m
e
n
tatio
n
t
e
c
h
n
iq
u
e
b
a
se
d
o
n
m
o
rp
h
o
l
o
g
ica
l
p
ro
c
e
ss
in
g
,
"
In
S
i
g
n
a
l
Pro
c
e
ss
in
g
a
n
d
In
fo
rm
a
t
io
n
T
e
c
h
n
o
lo
g
y
(
IS
S
PIT
),
IEE
E
I
n
ter
n
a
t
io
n
a
l
S
y
mp
o
siu
m
o
n
,
p
p
.
0
0
0
2
1
5
-
0
0
0
2
2
0
,
2
0
1
3
.
[1
8
]
D.
Da
s
a
n
d
S
.
M
u
k
h
o
p
a
d
h
y
a
y
,
"
F
in
g
e
rp
rin
t
Im
a
g
e
S
e
g
m
e
n
tatio
n
Us
in
g
Blo
c
k
-
Ba
se
d
S
tatisti
c
s
a
n
d
M
o
rp
h
o
l
o
g
ica
l
F
il
terin
g
,
"
Ara
b
.
J
.
S
c
i.
E
n
g
.,
v
o
l.
4
0
,
n
o
.
1
1
,
p
p
.
3
1
6
1
3
1
7
1
,
2
0
1
5
.
[1
9
]
A
.
A
,
A
b
b
o
o
d
,
S
.
G
h
a
z
a
li
,
A
.
R,
A
th
e
e
r
Ak
ra
m
a
n
d
U.P
.
S
a
b
i
n
e
,
"
S
e
g
m
e
n
tatio
n
a
n
d
e
n
h
a
n
c
e
m
e
n
t
o
f
f
in
g
e
rp
rin
t
im
a
g
e
s
b
a
se
d
o
n
a
u
to
m
a
ti
c
th
re
sh
o
ld
c
a
lcu
latio
n
s
,
"
In
In
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
f
Relia
b
le
In
fo
rm
a
t
io
n
a
n
d
Co
mm
u
n
ica
ti
o
n
T
e
c
h
n
o
l
o
g
y
S
p
ri
n
g
e
r
,
p
p
.
4
0
0
-
4
1
1
,
2
0
1
7
.
[2
0
]
V
.
M
h
a
sk
e
-
Dh
a
m
d
h
e
re
a
n
d
S
.
V
a
n
jale
,
"
A
n
o
v
e
l
a
p
p
ro
a
c
h
f
o
r
p
h
ish
i
n
g
e
m
a
il
s
re
a
l
ti
m
e
c
la
s
sif
ica
ti
o
n
u
sin
g
k
-
m
e
a
n
s a
lg
o
rit
h
m
,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
E
n
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
7
,
n
o
.
1
.
2
,
9
6
-
1
0
0
,
2
0
1
8
.
[2
1
]
G
.
G
a
n
a
n
d
M
.
K.
P
.
Ng
,
"
k
-
m
e
a
n
s
c
lu
ste
rin
g
w
it
h
o
u
tl
ier rem
o
v
a
l
,
"
Pa
tt
e
rn
Rec
o
g
n
it
io
n
L
e
tt
e
rs
,
9
0
,
p
p
.
8
-
14
,
2
0
1
7
.
[2
2
]
J.
Qin
,
W
.
F
u
,
H.
G
a
o
a
n
d
W
.
X
.
Z
h
e
n
g
,
"
Distrib
u
ted
K
-
M
e
a
n
s
A
l
g
o
rit
h
m
a
n
d
F
u
z
z
y
C
-
M
e
a
n
s
A
l
g
o
rit
h
m
f
o
r
S
e
n
so
r
Ne
tw
o
rk
s
B
a
se
d
o
n
M
u
lt
i
a
g
e
n
t
Co
n
se
n
su
s
T
h
e
o
ry
,
"
IEE
E
tra
n
sa
c
ti
o
n
s
o
n
c
y
b
e
rn
e
ti
c
s
,
4
7
(
3
),
p
p
.
7
7
2
-
7
8
3
,
2
0
1
7
.
[2
3
]
Ester,
M
a
rti
n
,
H.
P
Krie
g
e
l,
J.
S
a
n
d
e
r,
X
.
Xu
,
"
A
d
e
n
sity
-
b
a
se
d
a
l
g
o
rit
h
m
f
o
r
d
isc
o
v
e
rin
g
c
lu
ste
rs
in
larg
e
sp
a
ti
a
l
d
a
tab
a
se
s w
it
h
n
o
ise
,
"
In
K
d
d
,
v
o
l.
9
6
,
n
o
.
3
4
,
p
p
.
2
2
6
-
2
3
1
,
1
9
9
6
.
[2
4
]
R.
C.
G
o
n
z
a
lez
a
n
d
R.
E.
W
o
o
d
s,
"
Dig
ial
I
m
a
g
e
P
ro
c
e
ss
in
g
,
"
Pre
n
ti
c
e
Ha
ll
,
2
0
0
3
.
[2
5
]
D.
M
a
io
,
D.
M
a
lt
o
n
i,
R.
Ca
p
p
e
ll
i
e
t
a
l,
"
F
v
c
2
0
0
4
:
T
h
ir
d
f
in
g
e
rp
rin
t
v
e
rif
ic
a
ti
o
n
c
o
m
p
e
ti
ti
o
n
b
i
o
m
e
tri
c
a
u
th
e
n
ti
c
a
ti
o
n
,
"
S
p
rin
g
e
r B
e
rlin
He
id
e
lb
e
rg
,
p
p
.
1
-
7
,
2
0
0
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.