I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
,
p
p
.
1
2
0
1
~
1
2
0
8
I
SS
N:
2
5
02
-
4752,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ee
cs.
v
2
1
.
i
2
.
pp
1
2
0
1
-
120
8
1201
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Reducing
i
m
a
g
e s
ea
rch t
i
m
e
by i
mpro
v
ed BOV
W w
i
th
w
a
v
elet
deco
m
po
sitio
n
M
o
ha
m
m
e
d E
l A
m
in K
o
urt
i
ch
e
1
,
M
o
ha
m
m
e
d B
ela
dg
ha
m
2
,
A
bd
el
m
a
l
i
k
,
T
a
leb
-
A
h
med
3
1
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
,
L
a
b
o
ra
to
ry
o
f
T
I
T
,
T
a
h
ri
M
o
h
a
m
m
e
d
Un
iv
e
rsit
y
,
Be
c
h
a
r,
A
l
g
e
ri
a
2
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
,
L
a
b
o
ra
to
ry
o
f
T
IT
,
T
a
h
ri
M
o
h
a
m
m
e
d
Un
iv
e
rsit
y
,
Be
c
h
a
r,
A
l
g
e
ria
3
L
a
b
o
ra
to
ry
o
f
IEM
N DO
A
E.
U
M
R
CNRS
8
5
2
,
Un
iv
e
rsity
o
f
V
a
len
c
ien
n
e
s,
V
a
len
c
ien
n
e
s,
F
ra
n
c
e
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
1
5
,
2
0
2
0
R
ev
i
s
ed
A
u
g
22
,
2
0
20
A
cc
ep
ted
Sep
7
,
2
0
20
In
t
h
e
las
t
d
e
c
a
d
e
,
th
e
b
a
g
o
f
v
isu
a
l
w
o
rd
s
(BOV
W
)
h
a
s
b
e
e
n
u
se
d
w
id
e
l
y
in
im
a
g
e
c
las
si
f
ic
a
ti
o
n
,
im
a
g
e
re
tr
iev
a
l
a
n
d
h
a
s
si
g
n
if
ica
n
tl
y
i
m
p
ro
v
e
d
th
e
p
e
rf
o
r
m
a
n
c
e
o
f
CBIR
s
y
st
e
m
.
I
n
th
is
p
a
p
e
r
w
e
p
ro
p
o
se
a
n
e
w
m
e
th
o
d
t
o
e
n
h
a
n
c
e
BOV
W
u
sin
g
f
e
a
tu
re
s
o
b
tai
n
e
d
f
ro
m
w
a
v
e
let
d
e
c
o
m
p
o
siti
o
n
i
n
o
rd
e
r
t
o
r
e
d
u
c
e
c
o
m
p
u
tati
o
n
a
l
c
o
sts
in
v
o
c
a
b
u
lary
c
o
n
stru
c
ti
o
n
a
n
d
train
i
n
g
ti
m
e
.
W
e
a
p
p
l
y
se
v
e
ra
l
lev
e
l
o
f
w
a
v
e
let
d
e
c
o
m
p
o
siti
o
n
s
a
n
d
e
v
a
lu
a
te
th
e
ir
im
p
a
c
t
o
n
a
c
c
u
ra
c
y
o
f
th
e
BOV
W
.
W
e
a
p
p
l
y
o
u
r
m
e
th
o
d
o
n
M
URA
-
v
1
.
1
d
a
tas
e
t
a
n
d
t
h
e
e
x
p
e
rim
e
n
ts
re
su
lt
s co
n
f
irm
th
e
p
e
r
f
o
r
m
a
n
c
e
o
f
o
u
r
a
p
p
ro
a
c
h
.
K
ey
w
o
r
d
s
:
B
ag
o
f
v
i
s
u
al
w
o
r
d
s
I
m
ag
e
r
etr
ie
v
al
I
n
d
ex
i
n
g
W
av
elet
d
ec
o
m
p
o
s
itio
n
T
h
is
is
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Mo
h
a
m
m
ed
E
l
Am
in
Ko
u
r
tic
h
e
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
T
ah
r
i M
o
h
a
m
m
ed
U
n
i
v
er
s
it
y
I
n
d
ep
en
d
en
ce
r
o
ad
B
.
P
4
1
7
,
B
ec
h
ar
0
8
0
0
0
,
A
lg
er
ia
E
m
ail:
k
o
u
r
tic
h
e@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Du
e
to
tec
h
n
o
lo
g
ical
d
ev
e
lo
p
m
en
t
an
d
d
i
g
ital
eq
u
ip
m
e
n
t,
t
h
e
i
m
ag
e
h
a
s
b
ec
o
m
e
o
m
n
ip
r
esen
t
in
o
u
r
d
ail
y
li
v
es,
a
n
d
as
a
r
es
u
lt
a
h
u
d
g
e
m
as
s
o
f
i
m
a
g
es
ar
e
s
to
r
ed
ev
er
y
y
ea
r
,
it
is
a
v
er
y
d
i
f
f
i
cu
lt
ta
s
k
t
o
r
etr
ie
v
e
s
i
m
ilar
i
m
a
g
es
f
r
o
m
s
u
c
h
a
h
u
g
e
d
atab
ase.
T
h
e
f
ir
s
t
s
tu
d
ie
s
f
o
cu
s
ed
on
m
a
n
u
a
l
an
n
o
tati
o
n
w
h
ic
h
tak
e
s
a
lo
t
o
f
ti
m
e
an
d
r
eq
u
ir
es
a
lar
g
e
a
m
o
u
n
t
o
f
m
a
n
u
al
p
r
o
ce
s
s
i
n
g
i
n
ad
d
itio
n
,
it
s
till
s
u
b
j
ec
tiv
e
a
n
d
m
a
y
n
o
t
co
r
r
ec
tl
y
d
escr
ib
e
th
e
co
n
ten
t
o
f
th
e
i
m
ag
e
s
o
th
e
C
o
n
te
n
t
B
ased
I
m
a
g
e
R
etr
ie
v
al
s
y
s
te
m
s
(
CB
IR
)
h
av
e
ap
p
ea
r
ed
to
f
ill
th
e
g
ap
s
i
n
i
m
a
g
e
s
ea
r
ch
i
n
g
ar
ea
.
C
B
I
R
s
y
s
te
m
s
ar
e
b
ase
d
o
n
t
w
o
p
h
a
s
es
th
e
f
ir
s
t
i
s
ex
tr
ac
tio
n
o
f
i
m
a
g
e
f
ea
t
u
r
es
a
n
d
i
n
d
ex
i
n
g
,
t
h
e
s
ec
o
n
d
p
h
ase
is
s
ea
r
c
h
in
g
.
T
h
e
i
n
d
ex
i
n
g
tec
h
n
iq
u
e
s
ar
e
e
m
p
lo
y
ed
to
o
r
g
a
n
ize
a
n
d
ef
f
ec
tiv
e
l
y
r
ep
r
ese
n
t
t
h
e
co
n
ten
ts
o
f
a
d
atab
ase
.
A
s
th
e
i
m
ag
e
s
ar
e
in
ter
p
r
eted
as
in
d
ex
v
ec
to
r
s
,
th
e
n
th
e
s
ea
r
ch
is
d
o
n
e
in
t
h
e
in
d
ex
d
a
tab
ase.
T
h
e
s
ea
r
ch
p
r
o
ce
s
s
co
n
s
i
s
ts
o
f
s
ea
r
c
h
i
n
g
f
o
r
s
i
m
ilar
o
r
n
ea
r
v
ec
to
r
s
b
y
m
ea
s
u
r
in
g
t
h
e
d
i
s
tan
ce
b
et
w
e
en
t
h
e
t
w
o
v
ec
to
r
s
(
u
s
u
al
l
y
th
e
E
u
clid
ea
n
d
is
ta
n
ce
)
an
d
u
s
i
n
g
alg
o
r
it
h
m
s
s
u
c
h
as
k
-
Nea
r
es
t
Nei
g
h
b
o
r
.
Am
o
n
g
t
h
e
ap
p
r
o
ac
h
es
to
s
ea
r
ch
s
i
m
ilar
i
m
a
g
es
o
r
to
cla
s
s
i
f
y
a
n
i
m
a
g
e
to
t
h
eir
o
w
n
ca
teg
o
r
y
w
e
ca
n
cite
t
h
e
B
ag
o
f
v
is
u
al
w
o
r
d
s
(
B
OVW
)
.
B
OVW
h
as
r
ec
eiv
ed
m
u
c
h
a
tten
tio
n
in
t
h
e
p
ast
d
ec
ad
e
an
d
h
as
b
ee
n
r
e
v
is
ited
an
d
d
is
cu
s
s
ed
i
n
s
e
v
er
al
p
u
b
licatio
n
s
.
T
h
is
p
ap
er
p
r
esen
ts
a
n
e
w
ap
p
r
o
ac
h
t
o
i
m
p
r
o
v
e
th
e
p
er
f
o
r
m
a
n
ce
o
f
B
OVW
,
f
o
r
th
is
w
a
y
w
e
w
ill
u
s
e
t
h
e
f
ea
t
u
r
es
g
e
n
er
ated
b
y
w
a
v
ele
t
d
ec
o
m
p
o
s
itio
n
an
d
u
s
e
t
h
e
m
i
n
B
OVW
ap
p
r
o
ac
h
,
w
h
ic
h
g
i
v
e
u
s
o
u
tp
er
f
o
r
m
s
r
es
u
lt
s
in
ter
m
o
f
r
esp
o
n
s
e
ti
m
e
,
s
to
r
ag
e
ca
p
ac
it
y
.
T
h
e
o
r
g
an
izatio
n
o
f
th
is
p
ap
er
is
as
f
o
llo
w
s
:
Sectio
n
2
g
iv
e
s
a
b
r
ief
o
v
er
v
ie
w
o
f
r
elate
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
2
0
1
-
1
2
0
8
1202
w
o
r
k
t
h
en
Sectio
n
3
d
escr
ib
e
s
t
h
e
B
OVW
m
et
h
o
d
,
Sectio
n
4
a
n
al
y
s
e
s
w
a
v
elet
d
ec
o
m
p
o
s
itio
n
p
r
o
ce
s
s
,
o
u
r
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
o
u
t
lin
ed
in
t
h
e
Sec
tio
n
5
,
ex
p
er
i
m
en
ta
tio
n
a
n
d
r
es
u
lt
s
d
is
c
u
s
s
io
n
i
n
Sectio
n
6
,
a
n
d
o
u
r
co
n
clu
s
io
n
is
d
r
a
w
n
in
t
h
e
f
in
al
s
ec
tio
n
.
2.
RE
L
AT
E
D
WO
RK
Var
io
u
s
s
y
s
te
m
s
h
a
v
e
b
ee
n
i
n
tr
o
d
u
ce
d
f
o
r
C
B
I
R
s
y
s
te
m
s
u
s
in
g
t
h
e
ex
tr
ac
tio
n
o
f
lo
w
le
v
el
f
ea
t
u
r
es
o
f
i
m
ag
e,
s
u
ch
as
co
lo
r
,
s
h
ap
e
an
d
te
x
t
u
r
e
;
r
ec
en
t
tec
h
n
iq
u
e
s
ar
e
b
ased
o
n
k
e
y
p
o
in
t
s
e
x
tr
a
cted
f
r
o
m
r
eg
io
n
s
-
of
-
i
n
ter
est
o
f
th
e
i
m
a
g
es
w
h
i
ch
co
n
tain
r
ich
lo
ca
l
i
n
f
o
r
m
a
tio
n
ab
o
u
t
th
e
i
m
ag
e
[
1
]
,
t
h
is
k
e
y
p
o
i
n
ts
ca
n
b
e
d
etec
ted
u
s
in
g
d
i
f
f
er
e
n
t
d
ete
cto
r
s
[
2
]
an
d
r
ep
r
esen
ted
b
y
s
ev
er
al
d
escr
ip
to
r
s
[
3
]
.
Fo
r
in
d
ex
i
n
g
d
ata
an
d
m
i
n
i
m
ize
th
e
p
ath
to
r
etr
iev
e
an
i
m
ag
e
m
a
n
y
tr
ee
s
tr
u
ct
u
r
es
h
av
e
b
ee
n
i
n
tr
o
d
u
ce
d
s
u
c
h
as
R
-
tr
ee
,
R
*
-
tr
ee
,
an
d
SR
-
tr
ee
,
u
n
f
o
r
tu
n
atel
y
t
h
e
s
p
ee
d
an
d
ac
cu
r
ac
y
o
f
t
h
e
s
e
alg
o
r
ith
m
s
d
eg
r
ad
e
in
h
i
g
h
d
i
m
en
s
io
n
al
s
p
ac
e,
k
n
o
w
n
as
t
h
e
cu
r
s
e
o
f
d
i
m
en
s
i
o
n
alit
y
,
s
o
s
e
v
er
al
s
o
l
u
tio
n
s
h
av
e
r
ec
en
tl
y
b
ee
n
p
r
o
p
o
s
ed
to
s
o
lv
e
th
i
s
p
r
o
b
lem
s
u
c
h
as
:
R
ed
u
ctio
n
o
f
th
e
s
iz
e
s
u
c
h
as
P
C
A
(
p
r
in
cip
al
co
m
p
o
n
en
t
an
al
y
s
is
)
,
P
ar
allel
ar
ch
itect
u
r
e,
Hash
in
g
m
et
h
o
d
s
u
c
h
as
L
SH
(
L
o
ca
lit
y
Se
n
s
i
tiv
e
Has
h
in
g
)
,
B
ag
O
f
Vis
u
al
W
o
r
d
s
(
B
OVW
)
w
h
ic
h
w
e
d
escr
ib
e
it
i
n
d
etail
in
th
e
n
e
x
t sectio
n
.
B
ag
Of
Vis
u
al
W
o
r
d
s
h
as
b
ee
n
r
ev
i
s
ited
a
n
d
d
is
cu
s
s
ed
i
n
s
ev
er
al
p
u
b
licatio
n
s
,
in
[
1
]
t
h
e
au
th
o
r
s
s
tu
d
ied
v
ar
io
u
s
r
ep
r
ese
n
tatio
n
ch
o
ices
s
u
c
h
a
s
v
o
ca
b
u
lar
y
s
i
ze
,
w
ei
g
h
ti
n
g
,
w
o
r
d
s
elec
tio
n
an
d
th
e
ir
i
m
p
ac
t
to
class
i
f
icatio
n
p
er
f
o
r
m
a
n
ce
.
I
n
[
4
]
T
h
e
ex
p
er
im
e
n
t
s
co
n
f
ir
m
th
at
th
e
p
er
f
o
r
m
a
n
ce
o
f
a
B
O
V
W
s
y
s
te
m
ca
n
b
e
g
r
ea
tl
y
en
h
a
n
ce
d
b
y
tak
in
g
t
h
e
d
escr
i
p
to
r
s
s
p
atial
d
is
tr
ib
u
tio
n
i
n
to
ac
co
u
n
t
u
s
in
g
t
h
e
p
ar
titi
o
n
i
n
g
o
f
th
e
i
m
a
g
es
w
it
h
g
eo
m
etr
ic
tili
n
g
m
as
k
s
.
In
[
5
]
ex
p
lo
it
f
u
zz
y
cl
u
s
ter
i
n
g
f
o
r
co
d
eb
o
o
k
g
en
er
at
io
n
in
co
m
b
i
n
atio
n
w
it
h
s
o
f
t
as
s
i
g
n
m
e
n
ts
a
n
d
co
m
p
ar
ed
it
w
it
h
t
h
e
tr
ad
itio
n
al
co
d
eb
o
o
k
ap
p
r
o
ac
h
u
s
in
g
h
ar
d
ass
ig
n
m
en
t
.
I
n
[
6
]
w
o
r
k
,
p
r
o
p
o
s
ed
an
ap
p
r
o
ac
h
to
in
co
r
p
o
r
ate
s
p
atial
in
f
o
r
m
ati
o
n
in
t
h
e
B
OVW
o
n
e
x
p
licit
g
lo
b
al
r
elatio
n
s
h
ip
s
a
m
o
n
g
t
h
e
s
p
atial
p
o
s
itio
n
s
o
f
v
is
u
al
w
o
r
d
s
,
i
n
s
p
ir
ed
b
y
th
e
w
o
r
k
o
f
t
h
e
[
7
]
w
h
o
p
r
o
p
o
s
ed
s
p
atial
p
y
r
a
m
id
m
atc
h
(
SP
M)
.
In
[
8
]
p
r
o
p
o
s
ed
a
n
e
w
ap
p
r
o
ac
h
to
r
ev
is
ited
SDL
C
(
So
r
ted
Do
m
i
n
an
t
L
o
ca
l
C
o
lo
r
)
,
th
at
d
iv
id
es
t
h
e
i
m
ag
e
s
i
n
to
b
lo
ck
s
,
an
d
g
e
n
er
ates
a
te
x
t
u
al
s
i
g
n
atu
r
e
f
o
r
ea
ch
b
lo
ck
an
d
u
s
in
g
w
ei
g
h
ti
n
g
s
c
h
e
m
e
b
ased
o
n
th
e
f
r
eq
u
e
n
c
y
o
f
th
e
v
is
u
al
w
o
r
d
s
i
n
th
e
co
llect
io
n
.
T
o
en
h
an
ce
SD
L
C
a
n
o
t
he
r
n
e
w
S
-
B
OV
W
ma
pp
in
g
f
u
n
ctio
n
,
ca
l
led
S
o
r
ted
Do
m
i
n
a
n
t
L
o
ca
l
C
o
lo
r
an
d
T
ex
tu
r
e
(
SD
L
C
T
)
p
r
o
p
o
s
ed
in
[
9
]
,
th
i
s
tech
n
i
q
u
e
co
n
s
is
ts
o
f
a
co
m
b
i
n
atio
n
o
f
r
ep
r
esen
ta
tio
n
s
b
ase
d
o
n
b
o
th
co
lo
r
as
w
ell
a
s
te
x
t
u
r
e
in
f
o
r
m
a
ti
on
[
10
]
p
r
o
p
o
s
e
to
ad
d
s
e
m
an
t
ic
in
f
o
r
m
atio
n
u
s
in
g
eig
e
n
v
ec
to
r
s
o
f
th
e
i
m
ag
e
p
atc
h
es
i
n
o
r
d
er
to
ex
ten
d
t
h
e
B
OVW
r
ep
r
esen
tatio
n
f
o
r
i
m
a
g
e
r
etr
iev
al
[
11
]
P
r
o
p
o
s
ed
a
m
et
h
o
d
f
o
r
lear
n
in
g
w
e
ig
h
ti
n
g
s
ch
e
m
es
b
ased
o
n
Gen
etic
P
r
o
g
r
am
m
i
n
g
to
b
o
o
s
t th
e
p
er
f
o
r
m
a
n
ce
o
f
cla
s
s
i
f
icatio
n
m
o
d
els r
el
y
i
n
g
o
n
th
e
B
OVW
.
3.
B
AG
O
F
VIS
UA
L
WO
RD
S
Use
b
ag
o
f
v
is
u
al
w
o
r
d
s
(
B
OVW
)
in
i
m
a
g
e
r
etr
iev
al
ta
s
k
s
w
as
f
ir
s
t
p
r
o
p
o
s
ed
b
y
[
12
]
t
ak
in
g
in
s
p
ir
atio
n
f
r
o
m
t
h
e
ap
p
r
o
ac
h
o
f
b
a
g
s
o
f
w
o
r
d
s
i
n
d
ex
in
g
a
n
d
s
ea
r
c
h
in
g
o
f
te
x
t
u
al
in
f
o
r
m
atio
n
,
t
h
e
te
x
ts
ar
e
r
ep
r
esen
ted
b
y
s
ets
o
f
w
o
r
d
s
f
r
o
m
a
v
o
ca
b
u
lar
y
b
u
i
lt
f
r
o
m
t
h
e
co
r
p
u
s
,
t
h
er
ef
o
r
e
ev
er
y
te
x
t
is
r
ep
r
esen
ted
by
a
h
is
to
g
r
a
m
.
T
h
is
r
ep
r
esen
tatio
n
h
as
p
r
o
v
ed
to
b
e
v
er
y
ef
f
ec
ti
v
e
in
i
m
ag
e
clas
s
i
f
icatio
n
,
o
b
ject
r
ec
o
g
n
itio
n
an
d
o
b
j
ec
t d
etec
tio
n
.
Use
B
OVW
m
o
d
el
to
f
in
d
s
i
m
ilar
i
m
a
g
es,
t
o
r
etr
iev
e
o
r
to
ass
ig
n
t
h
e
m
to
th
eir
o
w
n
ca
te
g
o
r
y
ca
n
b
e
d
iv
id
ed
i
n
to
t
w
o
m
aj
o
r
p
ar
ts
.
T
h
e
f
ir
s
t
p
ar
t
is
t
h
e
i
m
ag
e
r
ep
r
esen
tat
io
n
a
n
d
t
h
e
s
ec
o
n
d
p
ar
t
co
n
s
is
t
to
tr
ain
a
k
er
n
e
l
m
et
h
o
d
.
As ill
u
s
tr
ated
in
Fig
u
r
e
1
t
h
e
r
ep
r
esen
tatio
n
p
ar
t c
an
b
e
d
iv
id
ed
in
to
th
r
e
e
s
tep
s
:
T
h
e
f
ir
s
t
s
tep
i
s
t
h
e
ex
tr
ac
t
i
o
n
o
f
f
ea
t
u
r
e
s
w
h
ic
h
is
t
h
e
r
e
s
u
lt
o
f
t
h
e
d
etec
tio
n
o
f
lo
ca
l
p
o
in
ts
o
f
in
ter
est
(
k
e
y
p
o
in
ts
)
an
d
th
e
n
d
escr
i
b
e
th
em
.
T
h
e
k
e
y
p
o
i
n
t
s
ca
n
b
e
d
etec
ted
b
y
v
ar
io
u
s
d
etec
to
r
s
[
2
]
an
d
d
escr
ib
ed
b
y
d
if
f
er
e
n
t
d
escr
ip
to
r
s
[
3
]
l
ik
e
SIFT
,
P
C
A
-
SIFT
,
SUR
F,
B
R
I
SK.
A
f
ter
ex
tr
ac
t
in
g
f
ea
t
u
r
e
v
ec
to
r
s
f
r
o
m
ea
c
h
i
m
a
g
e
i
n
th
e
d
ataset
,
th
e
n
ex
t
s
tep
i
s
t
o
co
n
s
tr
u
ct
t
h
e
d
ictio
n
ar
y
(
v
o
ca
b
u
lar
y
o
r
co
d
eb
o
o
k
)
,
w
h
ich
i
s
ac
h
iev
ed
b
y
c
lu
s
ter
in
g
t
h
e
f
e
atu
r
e
v
ec
to
r
s
o
b
tai
n
ed
f
r
o
m
all
i
m
a
g
es
o
f
d
atas
et
i
n
s
tep
1
,
u
s
in
g
cl
u
s
ter
in
g
alg
o
r
ith
m
as
k
-
m
ea
n
s
o
r
its
v
ar
ian
t.
E
ac
h
cl
u
s
ter
ce
n
ter
s
(
i
.
e,
ce
n
tr
o
id
s
)
ar
e
tr
ea
ted
a
s
a
v
i
s
u
a
l
w
o
r
d
o
f
th
e
d
ictio
n
ar
y
a
n
d
t
h
e
v
o
ca
b
u
lar
y
s
ize
i
s
t
h
e
n
u
m
b
er
o
f
c
lu
s
ter
s
.
T
h
e
last
s
tep
in
th
i
s
p
ar
t
is
v
ec
to
r
q
u
an
tizatio
n
,
u
s
ed
to
q
u
an
tify
a
n
d
r
ep
r
esen
t
ea
ch
i
m
a
g
e
i
n
t
h
e
d
ata
s
et
b
y
a
h
is
to
g
r
a
m
o
f
len
g
t
h
k
w
h
ic
h
r
ef
er
to
th
e
n
u
m
b
er
o
f
clu
s
ter
s
g
e
n
er
ated
f
r
o
m
k
-
m
ea
n
s
(
v
is
u
al
v
o
ca
b
u
lar
y
)
,
wh
er
e
e
ac
h
lo
ca
l
d
escr
ip
to
r
o
f
d
i
m
en
s
io
n
d
f
r
o
m
an
i
m
a
g
e
is
ass
i
g
n
ed
to
th
e
clo
s
est
ce
n
tr
o
id
,
an
d
th
e
i
-
th
v
al
u
e
in
th
e
h
is
to
g
r
a
m
is
t
h
e
f
r
eq
u
e
n
c
y
o
f
th
e
i
-
th
v
is
u
al
w
o
r
d
in
t
h
e
i
m
a
g
e.
T
he
h
i
s
to
g
r
a
m
s
w
o
r
k
as
an
i
n
d
ex
i
n
g
v
o
ca
b
u
lar
y
,
t
h
i
s
m
ea
n
s
t
h
at
th
e
B
OVW
r
ed
u
ce
s
t
h
e
s
iz
e
o
f
th
e
i
m
a
g
e
d
escr
ip
to
r
s
an
d
p
r
o
v
id
es
a
co
m
p
r
ess
ed
r
ep
r
esen
tatio
n
o
f
ea
c
h
i
m
a
g
e
in
th
e
d
ataset.
T
h
e
s
ec
o
n
d
p
ar
t
o
f
B
OVW
co
n
s
i
s
t
s
to
tr
ain
a
cla
s
s
i
f
ier
f
r
o
m
lab
eled
i
m
a
g
es
b
ased
o
n
t
h
e
r
ep
r
esen
tat
io
n
o
b
tain
ed
f
r
o
m
t
h
e
f
ir
s
t
p
ar
t
.
T
h
e
m
o
s
t
p
o
p
u
lar
class
if
ier
i
s
Su
p
p
o
r
t
V
ec
to
r
Ma
ch
in
e
s
(
SVM)
.
SV
M
is
f
le
x
ib
le
w
h
er
e
th
e
lear
n
in
g
k
er
n
el
can
be
v
ar
i
ed
ac
co
r
d
in
g
to
th
e
t
y
p
e
of
d
at
a
u
s
ed
[
13
,
14
]
s
u
ch
as:
li
n
ea
r
,
q
u
ad
r
atic,
R
ad
ial
B
asis
Fu
n
ctio
n
(
R
B
F),
χ
2
,
an
d
E
MD
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
R
ed
u
cin
g
ima
g
e
s
ea
r
ch
time
b
y
imp
r
o
ve
d
B
OV
W
w
i
th
…
(
M
o
h
a
mme
d
E
l A
min
K
o
u
r
tich
e
)
1203
Fig
u
r
e
1.
I
m
a
g
e
r
ep
r
esen
tat
io
n
u
s
i
n
g
b
a
g
o
f
v
is
u
al
w
o
r
d
s
4.
WAVE
L
E
T
AND
G
A
L
L
W
AVE
L
E
T
4
.
1
.
Wa
v
elet
deco
m
po
s
it
io
n
W
av
elet
d
ec
o
m
p
o
s
it
io
n
h
a
s
b
ee
n
w
id
el
y
u
s
ed
in
i
m
a
g
e
p
r
o
ce
s
s
in
g
i
n
v
ar
io
u
s
f
ie
ld
s
:
b
io
m
etr
ic
id
en
ti
f
icatio
n
,
co
m
p
r
e
s
s
io
n
,
class
i
f
icatio
n
,
i
m
ag
e
r
etr
ie
v
a
l
,
i
m
a
g
e
W
ater
m
ar
k
in
g
[
15
-
23
]
an
d
h
as
m
an
y
ad
v
an
ta
g
es
o
v
er
Fo
u
r
ier
tr
an
s
f
o
r
m
.
W
av
elet
tr
an
s
f
o
r
m
is
a
w
e
ll
lo
ca
lized
i
n
b
o
th
th
e
t
i
m
e
a
n
d
f
r
eq
u
en
c
y
d
o
m
ai
n
.
T
h
er
ef
o
r
e,
it
m
a
y
d
ec
o
m
p
o
s
e
a
s
i
g
n
al
r
etai
n
i
n
g
t
h
e
in
f
o
r
m
atio
n
o
f
b
o
th
d
o
m
ai
n
s
.
W
av
elet
tr
an
s
f
o
r
m
d
ec
o
m
p
o
s
e
s
a
s
i
g
n
al
w
i
th
a
f
a
m
il
y
o
f
b
asis
f
u
n
ctio
n
s
m
,
n
(
)
o
b
tain
ed
th
r
o
u
g
h
tr
an
s
latio
n
an
d
d
ilatio
n
[
24
]
o
f
a
m
o
th
er
w
a
v
elet.
m
,
n
(
)
=
(
)
w
h
er
e
m
a
n
d
n
ar
e
d
ilatio
n
an
d
tr
an
s
latio
n
p
ar
a
m
e
ter
s
.
T
h
e
w
a
v
elet
tr
an
s
f
o
r
m
a
tio
n
o
n
a
2
D
s
i
g
n
al
i
n
cl
u
d
es
r
ec
u
r
s
i
v
e
f
ilter
i
n
g
a
n
d
s
u
b
-
s
a
m
p
li
n
g
[
25
]
.
T
h
e
w
a
v
elet
tr
a
n
s
f
o
r
m
s
ca
n
b
e
co
m
p
u
ted
b
y
f
ir
s
t
p
er
f
o
r
m
i
n
g
1
D
DW
T
(
h
o
r
iz
o
n
tall
y
)
o
n
t
h
e
r
o
w
s
.
T
h
en
w
e
w
i
ll
do
th
e
s
a
m
e
1
D
DW
T
o
n
t
h
e
co
lu
m
n
s
(
v
er
ticall
y
)
f
o
r
b
o
th
t
h
e
lo
w
-
p
as
s
a
n
d
h
i
g
h
-
p
as
s
s
u
b
b
an
d
s
ig
n
al
s
o
b
tain
ed
f
r
o
m
th
e
h
o
r
izo
n
tal
a
n
al
y
s
i
s
A
t
ea
c
h
lev
el
t
h
e
s
i
g
n
al
is
d
ec
o
m
p
o
s
ed
i
n
to
f
o
u
r
f
r
eq
u
en
c
y
s
u
b
-
b
an
d
s
as
s
h
o
w
n
i
n
th
e
F
ig
u
r
e
(
a
)
,
L
L
ca
lled
ap
p
r
o
x
i
m
atio
n
,
L
H
k
n
o
w
n
a
s
v
er
tical
d
etail
s
,
H
L
ca
l
l
ed
h
o
r
izo
n
tal
d
etail
s
,
a
n
d
HH
k
n
o
w
n
a
s
d
iag
o
n
al
d
etails,
w
h
er
e
L
d
en
o
tes
th
e
l
o
w
f
r
eq
u
en
c
y
a
n
d
H
d
en
o
tes
th
e
h
ig
h
f
r
eq
u
en
c
y
.
Fo
r
th
e
n
ex
t
d
ec
o
m
p
o
s
itio
n
lev
el
(
i.e
s
ec
o
n
d
lev
el)
,
w
e
u
s
ed
th
is
s
a
m
e
p
r
o
ce
s
s
,
h
o
w
e
v
er
w
e
tak
e
t
h
e
ap
p
r
o
x
i
m
atio
n
s
u
b
-
i
m
a
g
e
of
t
h
e
p
r
ev
io
u
s
d
ec
o
m
p
o
s
itio
n
lev
el
(
i.e
f
ir
s
t le
v
e
l )
as sh
o
w
n
i
n
th
e
i
m
ag
e
Fig
u
r
e
(
b
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
2
0
1
-
1
2
0
8
1204
(
a
)
(
b
)
Fig
u
r
e
2
.
W
av
elet
d
ec
o
m
p
o
s
it
io
n
o
f
2
D
-
i
m
a
g
e
s
i
g
n
al
(
a
)
. 2
-
lev
el
w
av
e
let
d
ec
o
m
p
o
s
itio
n
.
(
b
)
.
DW
T
o
f
th
e
p
ep
p
er
s
im
a
g
e
o
v
er
t
w
o
lev
e
ls
4
.
2
.
B
io
rt
ho
g
o
na
l
wa
v
elet
s
B
io
r
th
o
g
o
n
al
w
a
v
elets
5
/3
ar
e
p
ar
t
o
f
th
e
f
a
m
il
y
o
f
s
y
m
m
etr
ic
b
io
r
th
o
g
o
n
al
w
a
v
elets
o
f
C
o
h
e
n
-
Dau
b
ec
h
ie
s
-
Feau
v
ea
u
(
C
DF)
.
T
h
ey
ar
e
s
o
ca
lled
b
ec
au
s
e
t
h
e
s
u
p
p
o
r
t
w
id
th
o
f
t
h
eir
lo
w
-
p
ass
f
ilter
s
,
d
etai
led
in
T
ab
le
1
,
is
p
=
5
s
am
p
les
f
o
r
an
aly
s
is
a
n
d
p
=3
f
o
r
s
y
n
th
e
s
i
s
.
I
n
ad
d
itio
n
,
th
ey
h
a
v
e
̃
ze
r
o
m
o
m
e
n
t
s
.
Du
e
to
th
e
ir
r
elativ
e
s
i
m
p
lici
t
y
a
n
d
th
e
s
y
m
m
etr
y
t
h
e
y
o
f
f
er
,
t
h
e
5
/3
w
a
v
elets
p
r
esen
ted
i
n
Fi
g
u
r
e
3
ar
e
u
s
ed
en
o
u
g
h
i
n
i
m
a
g
e
co
d
in
g
.
T
h
e
w
a
v
elets
o
f
t
h
is
f
a
m
il
y
ar
e
also
ca
lled
Gall
(
̃
)
,
w
h
er
e
N
d
en
o
te
s
th
e
n
u
m
b
er
o
f
n
u
ll
m
o
m
e
n
ts
o
f
t
h
e
a
n
al
y
s
is
w
a
v
elet
an
d
̃
its
eq
u
i
v
a
len
t
to
s
y
n
t
h
esi
s
.
As
f
o
r
Dau
b
ec
h
ies
w
a
v
elet
s
,
it
is
p
o
s
s
ib
le
to
s
h
o
w
t
h
at
Gall
w
a
v
elet
s
h
a
v
e
m
i
n
i
m
al
s
u
p
p
o
r
t
f
o
r
a
g
iv
e
n
n
u
m
b
er
o
f
n
u
ll
m
o
m
e
n
t
s
(
̃
).
T
ab
le
1
.
C
o
ef
f
icie
n
ts
o
f
s
y
m
m
etr
ical
i
m
p
u
l
s
e
r
esp
o
n
s
e
s
o
f
lo
w
-
p
ass
f
il
ter
s
an
al
y
s
is
[
]
an
d
s
y
n
th
e
s
i
s
̃
[
]
as
s
o
ciate
d
w
ith
Ga
ll
w
av
e
lets
5
/3
n
n
h
0
n
h
0
~
0
1
.
0
6
0
6
6
0
1
7
1
7
7
9
8
2
0
.
7
0
7
1
0
6
7
8
1
1
8
6
5
5
1
0
.
3
5
3
5
5
3
3
9
0
5
9
3
2
7
0
.
3
5
3
5
5
3
3
9
0
5
9
3
2
7
2
-
0
.
1
7
6
7
7
6
6
9
5
2
9
6
6
4
Fig
u
r
e
3
.
Gall
w
a
v
elet
5
/3
an
a
l
y
s
i
s
an
d
its
d
u
al
̃
.
5.
O
UR
AP
P
RO
ACH
Ou
r
o
b
j
ec
tiv
e
is
to
im
p
r
o
v
e
th
e
r
esp
o
n
s
e
ti
m
e
o
f
th
e
B
OVW
b
y
f
ast
i
n
d
ex
i
n
g
th
e
i
m
ag
e
s
co
n
tain
e
d
in
t
h
e
d
ataset
w
it
h
o
u
t
h
a
v
in
g
a
s
i
g
n
i
f
ica
n
t
d
ec
r
ea
s
e
i
n
ac
c
u
r
ac
y
w
h
e
n
s
e
a
r
c
h
i
n
g
th
e
co
n
v
e
n
ie
n
t
cla
s
s
f
o
r
a
q
u
er
y
i
m
a
g
e
,
to
t
h
is
en
d
o
u
r
co
n
tr
ib
u
tio
n
co
n
s
is
t
s
to
i
n
co
r
p
o
r
ate
th
e
Gall
w
av
ele
t
d
ec
o
m
p
o
s
itio
n
tec
h
n
iq
u
e
d
u
r
in
g
t
h
e
r
ep
r
esen
tat
io
n
p
h
a
s
e
an
d
ev
a
lu
ate
th
e
ir
i
m
p
ac
t
in
ac
cu
r
ac
y
a
n
d
r
esp
o
n
s
e
ti
m
e
,
Fig
u
r
e
2
ill
u
s
tr
ate
s
o
u
r
ap
p
r
o
ac
h
.
Fo
llo
w
i
n
g
is
t
h
e
d
etailed
w
o
r
k
i
n
g
p
r
o
ce
d
u
r
e
f
o
r
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
:
a)
T
h
e
i
m
a
g
es
o
f
d
ata
s
et
ar
e
p
ar
titi
o
n
ed
i
n
to
t
w
o
s
et
s
,
o
n
e
f
o
r
tr
ain
i
n
g
s
tep
a
n
d
th
e
o
th
er
f
o
r
test
i
n
g
.
Fo
r
each
i
m
ag
e
i
n
t
h
e
tr
ain
in
g
s
et
we
ap
p
l
y
t
h
e
w
a
v
elet
d
ec
o
m
p
o
s
itio
n
. W
e
tak
e
o
n
l
y
t
he
ap
p
r
o
x
i
m
at
io
n
p
ar
t
(
s
u
b
-
i
m
a
g
e
)
in
s
tead
of
tak
in
g
t
h
e
i
m
a
g
e
i
n
its
to
tali
t
y
.
b)
Featu
r
e
ex
tr
ac
tio
n
o
f
ea
c
h
s
u
b
-
i
m
ag
e
i
s
th
e
n
e
x
t step
,
an
d
is
p
er
f
o
r
m
ed
b
y
t
h
e
SU
R
F
m
et
h
o
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
R
ed
u
cin
g
ima
g
e
s
ea
r
ch
time
b
y
imp
r
o
ve
d
B
OV
W
w
i
th
…
(
M
o
h
a
mme
d
E
l A
min
K
o
u
r
tich
e
)
1205
c)
T
h
e
co
n
s
tr
u
ct
io
n
o
f
t
h
e
v
o
ca
b
u
lar
y
i
s
o
b
tai
n
ed
b
y
c
lu
s
ter
i
n
g
all
v
ec
to
r
s
o
f
p
r
ev
io
u
s
s
te
p
u
s
i
n
g
KN
N
alg
o
r
ith
m
,
th
e
ce
n
ter
o
f
ea
ch
clu
s
ter
is
ca
lled
a
v
is
u
al
w
o
r
d
an
d
th
e
co
m
b
in
atio
n
o
f
v
i
s
u
al
w
o
r
d
s
d
eter
m
in
e
s
th
e
d
ictio
n
ar
y
.
d)
Usi
n
g
t
h
e
f
r
eq
u
en
cie
s
o
f
ea
c
h
v
is
u
al
w
o
r
d
in
a
n
i
m
a
g
e
,
w
e
ca
n
r
ep
r
esen
t
e
v
er
y
i
m
ag
e
b
y
a
h
i
s
to
g
r
a
m
,
th
at
is
u
s
ed
to
tr
ain
t
h
e
clas
s
i
f
i
er
.
e)
T
r
ain
in
g
class
if
ier
u
s
in
g
m
u
lticlas
s
SVM:
SVM
is
o
n
e
o
f
th
e
m
o
s
t
w
id
ely
u
s
ed
class
if
icatio
n
m
o
d
els
in
th
e
m
ac
h
in
e
lear
n
in
g
ap
p
licatio
n
s
[
26
]
,
it
is
co
m
m
o
n
ly
u
s
ed
in
th
e
class
if
icatio
n
o
f
d
ata
esp
ec
ially
in
h
ig
h
d
im
en
s
io
n
al
f
ea
tu
r
e
s
p
ac
es.
I
n
B
OVW
,
th
e
SVM
class
if
ier
is
tr
ain
ed
u
s
in
g
h
is
to
g
r
am
s
o
f
tr
ain
in
g
im
ag
es.
SVM
is
u
s
ed
to
estab
lis
h
an
o
p
tim
al
h
y
p
er
p
lan
e
w
h
ich
s
ep
ar
ates
th
e
d
if
f
er
en
t
class
es
o
f
ex
am
p
les.
SVM
r
u
n
s
an
alg
o
r
ith
m
th
at
ass
is
ts
in
f
in
d
in
g
th
e
o
p
tim
al
h
y
p
er
p
lan
e
b
ased
o
n
th
e
tr
ain
in
g
d
ata.
T
h
e
o
p
tim
al
h
y
p
er
p
lan
e
is
ch
o
s
en
s
u
ch
th
at
th
e
d
is
tan
ce
o
f
th
e
h
y
p
er
p
lan
e
f
r
o
m
th
e
n
e
ar
est
d
ata
p
o
in
t
o
n
eith
er
s
id
e
is
m
ax
im
u
m
.
I
n
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
,
th
e
m
u
lticlas
s
SVM
class
if
ier
is
u
s
ed
w
ith
lin
ea
r
k
er
n
el
to
class
if
y
th
e
im
ag
es
d
ataset.
Giv
e
th
e
lab
eled
tr
ain
in
g
d
ata
to
th
e
SVM,
it
p
r
o
d
u
ce
s
an
o
p
tim
al
h
y
p
er
p
lan
e,
w
h
ich
th
en
allo
w
s
it
to
p
r
ed
ict
th
e
class
o
f
a
q
u
er
y
im
ag
e.
Fig
u
r
e
4.
B
ag
o
f
v
is
u
al
w
o
r
d
s
p
ip
elin
e
f
o
r
o
u
r
ap
p
r
o
ac
h
6.
O
UR
E
XP
E
R
I
M
E
NT
S
T
h
e
ex
p
er
i
m
en
t
s
w
er
e
p
er
f
o
r
m
ed
o
n
m
ed
ical
Xr
a
y
i
m
a
g
es
MU
R
A
-
v
1
.
1
,
MU
R
A
is
o
n
e
o
f
t
h
e
lar
g
es
t
p
u
b
lic
r
ad
io
g
r
ap
h
ic
i
m
a
g
e
d
at
asets
u
s
ed
in
i
m
a
g
e
r
etr
iev
a
l
t
ask
s
,
clas
s
i
f
ied
b
y
ca
te
g
o
r
y
ac
co
r
d
in
g
to
t
h
e
p
ar
ts
o
f
th
e
b
o
d
y
e
x
a
m
in
ed
(
Fi
n
g
er
,
Fo
r
ea
r
m
,
Han
d
,
S
h
o
u
ld
er
,
E
lb
o
w
,
Hu
m
er
u
s
,
W
r
is
t)
m
o
r
e
d
etails
o
n
th
is
to
p
ic
ca
n
b
e
f
o
u
n
d
i
n
[
27
]
.
W
e
ai
m
to
in
v
esti
g
ate
t
h
e
i
m
p
ac
t o
f
u
s
in
g
a
w
av
ele
t d
ec
o
m
p
o
s
itio
n
i
n
t
h
e
B
a
g
o
f
Vi
s
u
a
l
W
o
r
d
s
an
d
ev
al
u
ate
d
i
f
f
er
en
t
p
ar
a
m
e
ter
s
:
V
o
ca
b
u
lar
y
co
n
s
tr
u
ct
io
n
ti
m
e,
T
r
ain
in
g
ti
m
e,
S
to
r
ag
e
a
n
d
A
cc
u
r
ac
y
.
I
n
o
u
r
ex
p
er
i
m
en
t
s
w
e
ta
k
e
t
h
r
ee
ca
te
g
o
r
ie
s
(
E
l
b
o
w
,
Ha
n
d
,
Sh
o
u
ld
er
)
Fig
u
r
e
3
illu
s
tr
ates
s
o
m
e
s
a
m
p
les.
T
h
e
B
O
VW
is
f
lex
ib
le
w
h
er
ea
s
d
if
f
er
en
t
tec
h
n
iq
u
es
ca
n
b
e
u
s
ed
in
ea
c
h
o
f
it
s
p
ar
ts
,
s
o
i
n
o
u
r
ex
p
er
i
m
e
n
ts
,
w
e
c
h
o
s
e
SUR
F
m
et
h
o
d
to
ex
tr
ac
t
f
ea
tu
r
es
a
n
d
K
-
m
ea
n
s
f
o
r
clu
s
ter
in
g
t
h
e
m
,
w
e
ad
o
p
t
SVM
w
h
ic
h
is
a
m
o
n
g
t
h
e
m
o
s
t
w
i
d
ely
u
s
ed
clas
s
if
ier
s
f
o
r
B
OV
W
.
W
e
f
ix
ed
o
u
r
s
ize
v
o
ca
b
u
lar
y
o
n
5
0
0
v
is
u
al
w
o
r
d
s
,
f
o
r
ev
er
y
e
x
p
er
i
m
e
n
t
w
e
u
s
e
3
0
%
i
m
a
g
es
f
r
o
m
ea
ch
ca
te
g
o
r
y
to
tr
ain
t
h
e
al
g
o
r
ith
m
a
n
d
7
0
%
f
o
r
test
i
n
g
p
h
ase
.
T
h
e
ex
p
er
i
m
en
t
s
ar
e
te
s
ted
o
n
a
s
y
s
te
m
w
i
th
I
n
tel
(
R
)
C
o
r
e
(
T
M)
i5
-
7
3
0
0
U
C
P
U
2
.
6
GHz
a
n
d
8
GB
R
A
M
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
2
0
1
-
1
2
0
8
1206
Fig
u
r
e
5
.
Sa
m
p
le
i
m
ag
e
s
f
r
o
m
MU
R
A
d
ataset
6
.
1
.
Dis
cu
s
s
io
n
T
h
e
f
ir
s
t
s
et
o
f
an
al
y
s
e
s
in
v
e
s
tig
a
ted
b
et
w
ee
n
th
e
o
r
i
g
in
a
l
d
ataset
an
d
th
e
d
ec
o
m
p
o
s
ed
d
ataset
on
lev
el
1
is
s
h
o
w
n
i
n
Fi
g
u
r
e
4
,
w
e
o
b
s
er
v
e
al
m
o
s
t
t
h
e
s
a
m
e
v
alu
e
o
f
ac
c
u
r
ac
y
i
n
th
e
ca
s
e
o
f
3
0
0
0
,
4
5
0
0
,
6
0
0
0
i
m
a
g
es.
W
e
n
o
te
t
h
at
i
n
t
h
e
tr
a
d
itio
n
al
B
OVW
t
h
e
m
ac
h
i
n
e
c
an
n
o
t c
a
lcu
la
te
t
h
e
las
t t
w
o
ca
s
es 7
5
0
0
an
d
9
0
0
0
i
m
a
g
es
d
u
e
to
th
e
e
x
ce
ed
in
g
o
f
th
e
ca
p
ac
it
y
o
f
t
h
e
m
ac
h
i
n
e.
I
n
th
e
r
ep
r
esen
tatio
n
p
h
a
s
e
w
e
ad
d
th
e
ti
m
e
o
f
w
a
v
elet
d
ec
o
m
p
o
s
i
tio
n
w
h
ic
h
r
ep
r
esen
t
s
3
%
to
7
%
i
n
t
h
is
lev
el
(
1
)
co
m
p
ar
ed
to
t
h
e
to
talit
y
o
f
ti
m
e
r
ep
r
esen
tatio
n
i
n
le
v
el
1
,
Fi
g
u
r
e
5
s
h
o
w
s
a
s
ig
n
i
f
ica
n
t
d
i
f
f
er
en
ce
d
u
r
in
g
t
h
e
v
o
ca
b
u
lar
y
co
n
s
tr
u
ct
io
n
p
h
a
s
e.
E
rr
o
r!
Ref
er
ence
s
o
urce
n
o
t
f
o
un
d.
Fig
u
r
e
6
p
r
esen
ts
t
h
e
tr
ain
i
n
g
ti
m
e
,
w
e
n
o
te
th
at
th
e
ti
m
e
i
n
o
u
r
ap
p
r
o
ac
h
in
th
e
ca
s
e
o
f
9
0
0
0
i
s
lo
w
er
th
a
n
th
e
tr
ad
itio
n
al
B
OVW
in
th
e
ca
s
e
o
f
3
0
0
0
.
Fro
m
th
e
h
is
to
g
r
a
m
o
f
Fig
u
r
e
7
w
e
ca
n
s
ee
th
a
t
th
e
s
to
r
ag
e
ca
p
ac
it
y
o
f
d
ataset
o
f
o
u
r
ap
p
r
o
ac
h
r
ep
r
esen
ts
2
2
%
o
f
th
e
o
r
i
g
i
n
al
d
ataset.
Fig
u
r
e
6
.
A
cc
u
r
ac
y
Fig
u
r
e
7.
Vo
ca
b
u
lar
y
co
n
s
tr
u
c
tio
n
Fig
u
r
e
8.
Ker
n
el
tr
ain
i
n
g
Fig
u
r
e
9.
Sto
r
ag
e
Ou
r
tech
n
iq
u
e
s
h
o
w
s
clea
r
l
y
h
as
a
n
ad
v
a
n
ta
g
e
o
n
v
o
ca
b
u
la
r
y
ti
m
e
co
n
s
tr
u
ctio
n
an
d
tr
ai
n
in
g
ti
m
e
w
it
h
m
i
n
i
m
u
m
s
to
r
a
g
e
ca
p
ac
it
y
co
m
p
ar
ed
to
tr
ad
itio
n
al
o
n
e.
I
n
t
h
e
f
o
llo
w
i
n
g
s
t
u
d
ies
w
e
w
an
t
to
f
in
d
th
e
s
u
i
tab
le
d
ec
o
m
p
o
s
itio
n
lev
el
w
h
ic
h
d
o
es
n
o
t
af
f
ec
t
t
h
e
ac
cu
r
ac
y
r
ate
w
it
h
a
r
ea
s
o
n
ab
le
r
esp
o
n
s
e
ti
m
e,
f
o
r
th
i
s
w
e
v
ar
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
R
ed
u
cin
g
ima
g
e
s
ea
r
ch
time
b
y
imp
r
o
ve
d
B
OV
W
w
i
th
…
(
M
o
h
a
mme
d
E
l A
min
K
o
u
r
tich
e
)
1207
in
th
e
d
ec
o
m
p
o
s
itio
n
lev
el
an
d
th
e
m,
n
s
ize
o
f
d
ataset,
th
e
r
esu
lts
ar
e
s
h
o
w
n
b
elo
w
.
I
t
is
in
ter
esti
n
g
to
n
o
te
th
at
ea
c
h
ti
m
e
w
h
e
n
w
e
in
cr
e
ase
th
e
le
v
el
o
f
w
av
e
let
d
ec
o
m
p
o
s
itio
n
,
th
e
r
esp
o
n
s
e
a
n
d
t
r
ain
in
g
k
er
n
e
l
ti
m
e
s
as
w
e
ll a
s
t
h
e
s
to
r
ag
e
ca
p
ac
it
y
ar
e
r
ed
u
ce
d
. F
r
om
th
e
d
ictio
n
ar
y
co
n
s
tr
u
ctio
n
ti
m
e
as
s
h
o
wn
in
t
h
e
Fig
u
r
e
10
,
w
e
n
o
te
th
a
t
lev
e
l
1
co
n
s
u
m
es
m
o
r
e
co
m
p
u
tatio
n
ti
m
e
w
h
er
ea
s
le
v
el
2
d
o
es
n
o
t
ex
ce
ed
4
2
%
o
f
t
h
e
co
n
s
tr
u
ctio
n
ti
m
e
o
f
l
ev
el
1
a
n
d
is
w
ell
clea
r
l
y
s
h
o
w
n
i
n
all
o
t
h
er
ca
s
e
s
.
Fi
g
u
r
e
s
h
o
w
s
t
h
at
th
e
ac
cu
r
ac
y
at
lev
el
s
4
an
d
5
i
s
u
n
d
er
90
%,
an
d
it
is
alm
o
s
t
s
i
m
ilar
b
et
w
ee
n
lev
el
s
1
a
n
d
2
w
i
th
a
s
li
g
h
t
d
if
f
er
e
n
ce
at
lev
el
3
,
s
o
a
cc
o
r
d
in
g
to
th
e
g
r
ap
h
s
w
e
co
n
cl
u
d
e
th
at
th
e
d
ec
o
m
p
o
s
itio
n
at
lev
el
2
allo
w
ed
u
s
to
ac
h
iev
e
a
l
m
o
s
t
th
e
s
a
m
e
ac
c
u
r
ac
y
as
th
at
o
r
i
g
in
a
l
d
ataset
w
it
h
a
co
n
s
id
er
ab
le
g
ai
n
in
ti
m
e
a
n
d
s
to
r
ag
e
m
ed
iu
m
.
T
h
ese
te
s
ts
r
e
v
ea
led
th
at
w
a
v
elet
d
ec
o
m
p
o
s
i
tio
n
i
n
le
v
el
2
is
t
h
e
s
u
itab
le
s
o
lu
t
i
o
n
f
o
r
f
a
s
t
i
n
d
ex
i
n
g
an
d
tr
ain
i
n
g
B
OVW
.
K
er
n
el
tr
ain
i
n
g
a
n
d
Sto
r
ag
e
as s
h
o
w
n
i
n
Fi
g
u
r
e
1
1
an
d
1
2
.
Fig
u
r
e
10
.
V
o
ca
b
u
lar
y
co
n
s
tr
u
ctio
n
Fig
u
r
e
11
.
K
er
n
el
tr
ain
i
n
g
Fig
u
r
e
12.
Sto
r
ag
e
Fig
u
r
e
13.
A
cc
u
r
ac
y
7.
CO
NCLU
SI
O
N
I
n
th
is
w
o
r
k
,
w
e
p
r
o
p
o
s
e
a
n
e
w
m
eth
o
d
u
s
in
g
w
av
e
let
d
ec
o
m
p
o
s
i
tio
n
i
n
B
OVW
.
A
s
s
ta
ted
in
th
e
i
n
tr
o
d
u
ctio
n
o
u
r
m
a
in
ai
m
is
to
r
e
d
u
ce
th
e
ti
m
e
o
f
i
n
d
ex
in
g
to
r
etr
iev
e
an
d
s
ea
r
ch
r
e
lev
an
t
i
m
a
g
e
s
.
W
e
s
tu
d
ied
also
t
h
e
i
m
p
ac
t
o
f
t
h
e
w
a
v
elet
d
ec
o
m
p
o
s
itio
n
le
v
el,
o
u
r
m
et
h
o
d
allo
w
ed
B
O
VW
to
r
ed
u
ce
th
e
n
u
m
b
er
o
f
th
e
f
ea
t
u
r
e
s
v
ec
to
r
s
w
h
ic
h
i
m
p
ac
t
s
d
ir
ec
tl
y
o
n
t
h
e
co
m
p
u
ta
tio
n
al
co
s
t
i
n
B
OVW
an
d
a
s
ex
p
ec
ted
o
u
r
ex
p
er
i
m
en
t
s
p
r
o
v
e
th
at
u
s
e
w
av
e
le
t
d
ec
o
m
p
o
s
itio
n
i
n
B
OVW
p
ip
eli
n
e
h
av
e
a
s
i
g
n
i
f
ica
n
t
b
en
e
f
it
in
in
d
ex
i
n
g
ti
m
e,
k
er
n
e
l tr
ain
i
n
g
an
d
s
to
r
ag
e
ca
p
ac
it
y
.
RE
F
E
R
E
NC
E
S
[1
]
J.
Ya
n
g
,
Y.
-
G
.
Ji
a
n
g
,
A
.
G
.
Ha
u
p
tm
a
n
n
e
t
a
l.
,
“
Ev
a
lu
a
ti
n
g
b
a
g
-
of
-
v
isu
a
l
-
w
o
rd
s
re
p
re
s
e
n
tatio
n
s
in
sc
e
n
e
c
las
si
f
ica
ti
o
n
,
”
in
Pro
c
e
e
d
in
g
s
o
f
th
e
I
n
ter
n
a
t
io
n
a
l
W
o
rk
sh
o
p
o
n
W
o
rk
sh
o
p
o
n
M
u
lt
ime
d
i
a
I
n
f
o
rm
a
ti
o
n
R
e
triev
a
l
,
A
u
g
sb
u
rg
,
Ba
v
a
ri
a
,
Ge
r
m
a
n
y
,
p
p
.
1
9
7
-
2
0
6
,
2
0
0
7
.
[2
]
K.
M
ik
o
lajc
z
y
k
,
a
n
d
C.
S
c
h
m
id
,
“
S
c
a
le
&
Aff
in
e
In
v
a
rian
t
In
te
re
st
P
o
in
t
De
tec
to
rs,”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
mp
u
ter
V
isio
n
,
v
o
l.
6
0
,
n
o
.
1
,
p
p
.
6
3
-
86,
2
0
0
4
.
[3
]
K.
M
ik
o
lajc
z
y
k
,
a
n
d
C.
S
c
h
m
id
,
“
A
p
e
r
f
o
r
m
a
n
c
e
e
v
a
lu
a
ti
o
n
o
f
lo
c
a
l
d
e
sc
rip
to
rs,”
IEE
E
T
ra
n
s
Pa
tt
e
rn
An
a
l
M
a
c
h
In
tell,
v
o
l
.
2
7
,
n
o
.
1
0
,
p
p
.
1
6
1
5
-
1
6
3
0
,
2
0
0
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
21
,
No
.
2
,
Feb
r
u
ar
y
2
0
2
1
:
1
2
0
1
-
1
2
0
8
1208
[4
]
V
.
V
ii
tan
iem
i,
a
n
d
J.
L
a
a
k
so
n
e
n
,
"
S
p
a
ti
a
l
e
x
ten
sio
n
s
to
b
a
g
o
f
v
isu
a
l
w
o
rd
s
,
"
Pro
c
e
e
d
in
g
s
o
f
th
e
ACM
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Im
a
g
e
a
n
d
V
id
e
o
Retrie
v
a
l
,
p
p
.
1
-
8
,
2
0
0
9
.
[5
]
M
.
Ko
g
ler,
a
n
d
M
.
L
u
x
,
"
Ba
g
o
f
v
isu
a
l
w
o
rd
s
re
v
isit
e
d
:
a
n
e
x
p
lo
r
a
to
ry
stu
d
y
o
n
ro
b
u
st
im
a
g
e
re
tri
e
v
a
l
e
x
p
lo
it
in
g
f
u
z
z
y
c
o
d
e
b
o
o
k
s
,
"
Pro
c
e
e
d
in
g
s o
f
th
e
T
e
n
th
I
n
ter
n
a
ti
o
n
a
l
W
o
rk
sh
o
p
o
n
M
u
lt
ime
d
ia
D
a
ta
M
in
i
n
g
,
p
p
.
1
-
6
,
2
0
1
0
.
[6
]
R.
Kh
a
n
,
C.
Ba
ra
t,
D.
M
u
se
let
e
t
a
l.
,
"
S
p
a
ti
a
l
o
rien
tati
o
n
s
o
f
v
isu
a
l
w
o
rd
p
a
irs
to
im
p
ro
v
e
b
a
g
-
of
-
v
isu
a
l
-
w
o
rd
s
m
o
d
e
l
,
"
in
:
Brit
ish
M
a
c
h
in
e
V
isio
n
Co
n
fer
e
n
c
e
,
BM
VA
,
2
0
1
2
.
[7
]
S
.
L
a
z
e
b
n
ik
,
C.
S
c
h
m
id
,
a
n
d
J.
P
o
n
c
e
,
"
Be
y
o
n
d
b
a
g
s
o
f
f
e
a
tu
re
s:
S
p
a
ti
a
l
p
y
ra
m
id
m
a
tch
in
g
f
o
r
r
e
c
o
g
n
izin
g
n
a
tu
ra
l
sc
e
n
e
c
a
te
g
o
ries
,
"
IEE
E
Co
mp
u
t
e
r
S
o
c
ie
ty
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
sio
n
a
n
d
Pa
t
ter
n
Rec
o
g
n
i
ti
o
n
(
CVP
R
'
0
6
)
,
v
o
l.
2
,
p
p
.
2
1
6
9
-
2
1
7
8
,
2
0
0
6
.
[8
]
J.
M
.
d
o
s
S
a
n
t
o
s,
E.
S
.
d
e
M
o
u
ra
,
A
.
S
.
d
a
S
il
v
a
e
t
a
l.
,
“
A
sig
n
a
tu
re
-
b
a
se
d
b
a
g
o
f
v
isu
a
l
w
o
rd
s
m
e
t
h
o
d
f
o
r
im
a
g
e
in
d
e
x
in
g
a
n
d
se
a
rc
h
,
”
Pa
tt
e
rn
Re
c
o
g
n
i
t
L
e
tt
,
v
o
l.
6
5
,
p
p
.
1
-
7
,
2
0
1
5
.
[9
]
J.
M
.
Do
s
S
a
n
to
s,
E.
S
.
De
M
o
u
ra
,
A
.
S
.
Da
S
il
v
a
e
t
a
l.
,
“
Co
lo
r
a
n
d
tex
tu
re
a
p
p
li
e
d
to
a
sig
n
a
tu
re
-
b
a
se
d
b
a
g
o
f
v
isu
a
l
w
o
rd
s
m
e
th
o
d
f
o
r
im
a
g
e
r
e
tri
e
v
a
l,
”
M
u
lt
ime
d
ia
T
o
o
ls
a
n
d
Ap
p
li
c
a
ti
o
n
s,
v
o
l.
7
6
,
n
o
.
1
5
,
p
p
.
1
6
8
5
5
-
1
6
8
7
2
,
2
0
1
7
.
[1
0
]
N.
Bh
a
tt
a
c
h
a
r
y
a
,
a
n
d
J.
S
il
,
"
I
m
a
g
e
r
e
tri
e
v
a
l
u
sin
g
e
x
ten
d
e
d
b
a
g
-
of
-
v
isu
a
l
-
w
o
rd
s
,
"
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Ad
v
a
n
c
e
s i
n
Co
m
p
u
ti
n
g
,
Co
mm
u
n
ica
ti
o
n
s
a
n
d
In
fo
rm
a
t
ics
(
ICACCI)
,
p
p
.
1
9
6
9
-
1
9
7
5
,
2
0
1
6
.
[1
1
]
R.
T
.
Io
n
e
sc
u
,
a
n
d
M
.
P
o
p
e
sc
u
,
"
Ob
jec
t
re
c
o
g
n
it
io
n
w
it
h
t
h
e
b
a
g
o
f
v
isu
a
l
w
o
rd
s
m
o
d
e
l
,
"
Kn
o
w
led
g
e
T
r
a
n
sfe
r
b
e
twee
n
Co
mp
u
ter
Vi
si
o
n
a
n
d
T
e
x
t
M
in
in
g
:
S
imil
a
rity
-
b
a
se
d
L
e
a
rn
in
g
Ap
p
ro
a
c
h
e
s
,
p
p
.
9
9
-
1
3
2
,
Ch
a
m
:
S
p
rin
g
e
r
In
tern
a
ti
o
n
a
l
P
u
b
l
ish
i
n
g
,
p
p
.
9
9
-
1
3
2
,
2
0
1
6
.
[1
2
]
J.
S
iv
ic,
a
n
d
A
.
Zi
ss
e
r
m
a
n
,
"
V
id
e
o
G
o
o
g
le:
A
tex
t
r
e
tri
e
v
a
l
a
p
p
ro
a
c
h
to
o
b
jec
t
m
a
tch
in
g
in
v
id
e
o
s,"
Pro
c
e
e
d
in
g
s
o
f
th
e
Nin
t
h
IE
EE
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
Vi
si
o
n
,
p
p
.
1
4
7
0
-
1
4
7
8
,
2
0
0
3
.
[1
3
]
J.
Zh
a
n
g
,
M
.
M
a
rsz
a
łek
,
S
.
L
a
z
e
b
n
ik
e
t
a
l.
,
“
L
o
c
a
l
fe
a
tu
re
s
a
n
d
k
e
rn
e
ls
f
o
r
c
las
sif
i
c
a
ti
o
n
o
f
tex
tu
re
a
n
d
o
b
jec
t
c
a
teg
o
ries
:
A
c
o
m
p
re
h
e
n
siv
e
stu
d
y
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
m
p
u
ter
Vi
si
o
n
,
v
o
l.
7
3
,
n
o
.
2
,
p
p
.
2
1
3
-
2
3
8
,
2
0
0
7
.
[1
4
]
Y.
-
G
.
Jia
n
g
,
C.
-
W
.
Ng
o
,
a
n
d
J.
Ya
n
g
,
"
T
o
w
a
rd
s
o
p
ti
m
a
l
b
a
g
-
of
-
f
e
a
tu
re
s
f
o
r
o
b
jec
t
c
a
teg
o
riza
ti
o
n
a
n
d
se
m
a
n
ti
c
v
id
e
o
re
tri
e
v
a
l
,
"
Pro
c
e
e
d
in
g
s
o
f
t
h
e
6
t
h
ACM
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
Ima
g
e
a
n
d
v
i
d
e
o
re
triev
a
l,
p
p
.
4
9
4
-
5
0
1
,
2
0
0
7
.
[1
5
]
Y.
W
a
n
g
,
X
.
Din
g
,
R.
W
a
n
g
e
t
a
l.
,
"
F
u
si
o
n
-
b
a
se
d
u
n
d
e
rw
a
ter
ima
g
e
e
n
h
a
n
c
e
m
e
n
t
b
y
w
a
v
e
let
d
e
c
o
m
p
o
siti
o
n
,
"
IEE
E
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
I
n
d
u
stri
a
l
T
e
c
h
n
o
l
o
g
y
(
ICIT
)
,
p
p
.
1
0
1
3
-
1
0
1
8
,
2
0
1
7
.
[1
6
]
A
.
B.
S
a
id
,
I.
Je
m
e
l,
R.
Ej
b
a
li
e
t
a
l.
,
"
A
h
y
b
rid
a
p
p
r
o
a
c
h
f
o
r
i
m
a
g
e
c
las
si
f
ica
ti
o
n
b
a
se
d
o
n
s
p
a
rse
c
o
d
in
g
a
n
d
w
a
v
e
let
d
e
c
o
m
p
o
siti
o
n
,
"
IEE
E/
ACS
1
4
t
h
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
S
y
ste
ms
a
n
d
Ap
p
l
ica
ti
o
n
s
(
AICCS
A)
,
p
p
.
6
3
-
68
,
2
0
1
7
.
[1
7
]
N.
Qa
z
i,
a
n
d
B.
W
.
W
o
n
g
,
"
S
e
m
a
n
ti
c
b
a
se
d
im
a
g
e
r
e
tri
e
v
a
l
th
ro
u
g
h
c
o
m
b
in
e
d
c
las
si
f
iers
o
f
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
a
n
d
w
a
v
e
l
e
t
d
e
c
o
m
p
o
siti
o
n
o
f
ima
g
e
si
g
n
a
l
,
"
Pro
c
e
e
d
in
g
s
o
f
t
h
e
9
th
EUROS
IM
Co
n
g
re
ss
o
n
M
o
d
e
ll
in
g
a
n
d
S
imu
l
a
ti
o
n
,
EUROS
IM
,
n
o
.
1
4
2
,
p
p
.
4
7
3
-
4
7
8
,
2
0
1
6
.
[1
8
]
P
.
A
.
Ha
g
a
r
g
i,
a
n
d
D.
S
h
u
b
h
a
n
g
i,
“
Bra
in
t
u
m
o
r
M
R
im
a
g
e
f
u
sio
n
u
si
n
g
m
o
st
d
o
m
in
a
n
t
f
e
a
tu
re
s
e
x
trac
ti
o
n
f
ro
m
w
a
v
e
let
a
n
d
c
u
rv
e
let
tran
sf
o
r
m
s,”
Bra
in
,
v
o
l.
5
,
n
o
.
0
5
,
p
p
.
3
3
-
3
8
,
2
0
1
8
.
[1
9
]
A
.
S
e
t
y
o
n
o
,
a
n
d
D.
S
e
ti
a
d
i,
“
Im
a
g
e
W
a
ter
m
a
rk
in
g
u
sin
g
Disc
re
t
e
W
a
v
e
let
-
T
c
h
e
b
ich
e
f
T
ra
n
s
f
o
r
m
,
”
In
d
o
n
e
s.
J
.
El
e
c
tr.
En
g
.
Co
m
p
u
t
.
S
c
i.
,
v
o
l.
1
6
,
n
o
.
3
,
p
p
.
1
6
-
2
1
,
2
0
1
9
.
[2
0
]
A
.
V
a
ish
,
S
.
G
a
u
tam
,
a
n
d
M
.
Ku
m
a
r,
“
A
w
a
v
e
let
b
a
se
d
a
p
p
ro
a
c
h
f
o
r
sim
u
lt
a
n
e
o
u
s
c
o
m
p
re
ss
io
n
a
n
d
e
n
c
ry
p
ti
o
n
o
f
f
u
se
d
i
m
a
g
e
s,”
J
o
u
rn
a
l
o
f
Kin
g
S
a
u
d
U
n
ive
rs
it
y
-
Co
mp
u
ter
a
n
d
In
f
o
rm
a
ti
o
n
S
c
ien
c
e
s,
v
o
l.
3
1
,
n
o
.
2
,
p
p
.
2
0
8
-
2
1
7
,
2
0
1
9
.
[2
1
]
C.
Qin
,
Q.
Zh
o
u
,
F
.
Ca
o
e
t
a
l.
,
“
F
lex
ib
le
lo
ss
y
c
o
m
p
re
ss
io
n
f
o
r
se
lec
ti
v
e
e
n
c
r
y
p
ted
i
m
a
g
e
w
it
h
im
a
g
e
in
p
a
in
ti
n
g
,
”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Circ
u
it
s
a
n
d
S
y
ste
ms
fo
r V
id
e
o
T
e
c
h
n
o
lo
g
y
,
v
o
l.
2
9
,
n
o
.
1
1
,
p
p
.
3
3
4
1
-
3
3
5
5
,
2
0
1
8
.
[2
2
]
Z.
Ye
,
H.
M
o
h
a
m
a
d
ian
,
a
n
d
Y.
Ye
,
"
In
f
o
r
m
a
ti
o
n
m
e
a
su
re
s
f
o
r
b
io
m
e
tri
c
id
e
n
ti
f
ica
ti
o
n
v
ia
2
d
d
isc
re
te
w
a
v
e
let
tran
sf
o
r
m
,
"
IEE
E
In
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
Au
to
ma
ti
o
n
S
c
ien
c
e
a
n
d
E
n
g
i
n
e
e
rin
g
,
p
p
.
8
3
5
-
8
4
0
,
2
0
0
7
.
[2
3
]
P
.
S
riv
a
sta
v
a
,
a
n
d
A
.
Kh
a
re
,
“
In
teg
ra
ti
o
n
o
f
wa
v
e
let
tran
s
f
o
r
m
,
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s
a
n
d
m
o
m
e
n
ts
f
o
r
c
o
n
ten
t
-
b
a
se
d
im
a
g
e
re
tri
e
v
a
l,
”
J
o
u
rn
a
l
o
f
Vi
su
a
l
Co
mm
u
n
ica
t
io
n
a
n
d
Ima
g
e
Rep
re
se
n
ta
ti
o
n
,
v
o
l.
4
2
,
p
p
.
7
8
-
1
0
3
,
2
0
1
7
.
[2
4
]
M
.
A
n
to
n
i
n
i,
a
n
d
M
.
Ba
rlau
d
,
“
Im
a
g
e
c
o
d
in
g
u
si
n
g
w
a
v
e
let
tran
s
f
o
r
m
,
"
IEE
E
T
ra
n
sa
c
ti
o
n
o
n
Im
a
g
e
Pro
c
e
ss
in
g
,
v
o
l.
1
,
n
o
.
2
,
p
p
.
2
0
5
-
2
2
0
,
1
9
9
2
.
[2
5
]
C.
-
W
.
Ko
k
,
a
n
d
W
.
-
S
.
T
a
m
,
Dig
it
a
l
Ima
g
e
In
ter
p
o
l
a
ti
o
n
i
n
M
a
tl
a
b
:
Jo
h
n
W
il
e
y
&
S
o
n
s,
2
0
1
9
.
[2
6
]
O.
Ch
a
p
e
ll
e
,
P
.
Ha
f
f
n
e
r,
a
n
d
V
.
N.
V
a
p
n
ik
,
“
S
u
p
p
o
r
t
v
e
c
to
r
m
a
c
h
in
e
s
f
o
r
h
isto
g
ra
m
-
b
a
se
d
i
m
a
g
e
c
las
si
f
ica
ti
o
n
,
”
IEE
E
T
ra
n
s Ne
u
ra
l
Ne
tw,
v
o
l
.
1
0
,
n
o
.
5
,
p
p
.
1
0
5
5
-
1
0
6
4
,
1
9
9
9
.
[2
7
]
P
.
Ra
jp
u
rk
a
r,
J.
Irv
in
,
A
.
Ba
g
u
l
e
t
a
l.
,
“
M
u
ra
:
L
a
rg
e
d
a
tas
e
t
f
o
r
a
b
n
o
rm
a
li
ty
d
e
tec
ti
o
n
in
m
u
sc
u
lo
sk
e
leta
l
ra
d
io
g
ra
p
h
s,”
a
rXiv p
re
p
ri
n
t
a
rXi
v
:1
7
1
2
.
0
6
9
5
7
,
2
0
1
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.