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.
7
,
No
.
5
,
Octo
b
e
r
2
0
1
7
,
p
p
.
2
5
0
2
~2
51
3
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v7
i
5
.
pp
2
5
0
2
-
2
51
3
2502
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
JE
C
E
A Nov
el App
ro
a
ch Ba
sed o
n De
cre
a
sed Di
m
ensio
n a
nd Redu
ced
Gr
a
y
Lev
el Rang
e Ma
trix Feat
ures
f
o
r St
o
n
e Tex
tur
e
Cla
ss
ificatio
n
G
.
S.
N.
M
urt
hy
1
,
Srini
nv
a
s
a
Ra
o
.
V
2
,
T
.
Vee
rr
a
j
u
3
1
Ra
y
a
las
e
e
m
a
Un
iv
e
rsit
y
,
Ku
rn
o
o
l,
In
d
ia
2
De
p
t
o
f
CS
E,
V
.
R.
S
i
d
d
h
a
rth
a
E
n
g
in
e
e
r
in
g
Co
ll
e
g
e
,
V
ij
a
y
a
wa
d
a
,
In
d
ia
3
De
p
t
o
f
CS
E,
A
d
it
y
a
c
o
ll
e
g
e
o
f
En
g
in
e
e
rin
g
,
S
u
ra
m
p
a
le
m
,
In
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
3
1
,
2
0
1
7
R
ev
i
s
ed
J
u
n
5
,
2
0
1
7
A
cc
ep
ted
Sep
t
11
,
2
0
1
7
T
h
e
h
u
m
a
n
e
y
e
c
a
n
e
a
sil
y
id
e
n
ti
fy
th
e
t
y
p
e
o
f
tex
tu
r
e
s
in
f
lo
o
rin
g
o
f
th
e
h
o
u
se
s
a
n
d
in
t
h
e
d
ig
it
a
l
im
a
g
e
s
v
isu
a
ll
y
.
In
th
is
w
o
rk
,
th
e
sto
n
e
t
e
x
tu
re
s
a
re
g
ro
u
p
e
d
i
n
to
f
o
u
r
c
a
teg
o
ries
.
T
h
e
y
a
re
b
rick
s,
m
a
rb
le,
g
ra
n
it
e
a
n
d
m
o
sa
ic.
A
n
o
v
e
l
a
p
p
r
o
a
c
h
is
d
e
v
e
lo
p
e
d
f
o
r
d
e
c
re
a
sin
g
th
e
d
im
e
n
sio
n
o
f
sto
n
e
i
m
a
g
e
a
n
d
f
o
r
re
d
u
c
in
g
th
e
g
ra
y
le
v
e
l
ra
n
g
e
o
f
th
e
ima
g
e
w
it
h
o
u
t
a
n
y
lo
ss
o
f
sig
n
if
ica
n
t
f
e
a
tu
re
in
f
o
rm
a
ti
o
n
.
T
h
is
m
o
d
e
l
is
n
a
m
e
d
a
s
“
De
c
re
a
se
d
Di
m
e
n
sio
n
a
n
d
Re
d
u
c
e
d
G
ra
y
lev
e
l
Ra
n
g
e
M
a
tri
x
(DD
R
G
RM
)”
m
o
d
e
l.
T
h
e
DD
R
G
RM
m
o
d
e
l
c
o
n
sists
o
f
3
sta
g
e
s.
In
sta
g
e
1
,
e
a
c
h
5
×
5
su
b
d
im
e
n
sio
n
o
f
th
e
sto
n
e
im
a
g
e
is
re
d
u
c
e
d
in
to
2
×
2
s
u
b
d
im
e
n
sio
n
w
it
h
o
u
t
l
o
sin
g
a
n
y
im
p
o
rtan
t
q
u
a
li
ti
e
s,
p
rim
it
iv
e
s,
a
n
d
a
n
y
o
th
e
r
lo
c
a
l
stu
f
f
.
In
sta
g
e
2
,
th
e
g
ra
y
lev
e
l
o
f
th
e
im
a
g
e
is
re
d
u
c
e
d
f
ro
m
0
-
2
5
5
to
0
-
4
b
y
u
sin
g
f
u
z
z
y
c
o
n
c
e
p
ts.
In
sta
g
e
3
,
Co
-
o
c
c
u
rre
n
c
e
M
a
tri
x
(CM
)
fe
a
tu
re
s
a
re
d
e
riv
e
d
f
ro
m
th
e
DD
R
G
RM
m
o
d
e
l
o
f
th
e
sto
n
e
im
a
g
e
f
o
r
sto
n
e
tex
tu
re
c
las
si
f
ic
a
ti
o
n
.
Ba
se
d
o
n
th
e
f
e
a
tu
re
se
t
v
a
lu
e
s,
a
u
se
r
d
e
f
in
e
d
a
lg
o
rit
h
m
is
d
e
v
e
lo
p
e
d
to
c
las
sify
th
e
sto
n
e
tex
tu
re
im
a
g
e
in
to
o
n
e
o
f
th
e
4
c
a
teg
o
ries
i.
e
.
M
a
rb
le,
Brick
,
G
ra
n
it
e
a
n
d
M
o
sa
ic.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
is
tes
ted
b
y
u
sin
g
th
e
K
-
N
e
a
re
st
Ne
ig
h
b
o
r
Clas
sif
ic
a
ti
o
n
a
lg
o
rit
h
m
w
it
h
th
e
d
e
riv
e
d
tex
tu
re
f
e
a
tu
re
s.
T
o
p
ro
v
e
th
e
e
ff
ici
e
n
c
y
o
f
th
e
p
ro
p
o
se
d
m
e
th
o
d
,
it
is
tes
ted
o
n
d
if
fe
re
n
t
sto
n
e
tex
tu
re
im
a
g
e
d
a
tab
a
s
e
s.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
re
su
lt
e
d
in
h
ig
h
c
las
sif
i
c
a
ti
o
n
ra
te
w
h
e
n
c
o
m
p
a
re
d
w
it
h
th
e
o
th
e
r
e
x
isti
n
g
m
e
th
o
d
s.
K
ey
w
o
r
d
s
:
C
las
s
i
f
icatio
n
Co
-
o
cc
u
r
r
en
ce
m
atr
i
x
Fu
zz
y
l
o
g
ic
R
ed
u
ce
d
d
i
m
e
n
s
io
n
alit
y
S
to
n
e
tex
tu
r
e
Co
p
y
rig
h
t
©
2
0
1
7
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
:
G
.
S
.
N
.
Mu
r
th
y
,
A
d
it
y
a
C
o
lle
g
e
o
f
E
n
g
i
n
ee
r
i
n
g
,
Su
r
a
m
p
ale
m
,
An
d
h
r
a
P
r
ad
esh
,
I
n
d
ia
-
5
3
3
4
3
7
.
E
m
ail:
m
u
r
t
h
y
g
s
n
m
@
y
ah
o
o
.
c
o
m
1.
I
NT
RO
D
UCT
I
O
N
On
e
o
f
th
e
m
ain
co
n
te
n
t
s
o
f
th
e
i
m
a
g
e
is
tex
t
u
r
e
.
T
h
e
an
al
y
s
i
s
o
f
tex
t
u
r
es
m
ai
n
l
y
i
n
cl
u
d
es:
P
atter
n
s
id
en
ti
f
ied
o
n
te
x
t
u
r
e,
s
h
ap
es
r
ec
o
v
er
y
f
r
o
m
te
x
tu
r
e,
T
ex
t
u
r
e
s
e
g
m
e
n
tat
io
n
,
a
n
d
tex
t
u
r
e
class
i
f
icatio
n
[
1
].
Am
o
n
g
t
h
e
m
,
te
x
t
u
r
e
c
lass
i
f
ic
atio
n
p
la
y
s
a
n
i
m
p
o
r
tan
t
r
o
le
in
m
a
n
y
ar
ea
s
s
u
c
h
as
r
e
m
o
te
s
en
s
in
g
,
s
to
n
e
t
y
p
e
id
en
ti
f
icatio
n
in
co
n
s
tr
u
ctio
n
f
iled
,
m
ed
ical
i
m
a
g
i
n
g
an
d
s
o
o
n
[
2
]
.
T
ex
tu
r
e
a
n
al
y
s
is
i
s
o
n
e
o
f
t
h
e
p
r
i
m
e
tech
n
iq
u
es
w
h
e
n
ap
p
lied
o
n
im
ag
e
s
co
n
s
is
ti
n
g
o
f
r
ep
etitio
n
o
r
q
u
asi
r
ep
etitio
n
o
f
s
o
m
e
f
u
n
d
a
m
e
n
tal
i
m
a
g
e
ele
m
e
n
ts
.
T
h
e
y
ar
e
an
al
y
ze
d
an
d
in
ter
p
r
eted
b
y
Yag
n
a
n
ar
ay
an
a
[
3
]
.
Fro
m
t
h
e
p
ast
t
w
o
d
ec
ad
es,
s
o
m
an
y
tech
n
iq
u
es
h
av
e
b
ee
n
id
e
n
ti
f
i
ed
f
o
r
tex
tu
r
e
an
al
y
s
i
s
esp
ec
i
all
y
in
f
ea
t
u
r
e
ex
tr
ac
tio
n
a
n
d
class
i
f
icatio
n
ar
ea
s
.
T
h
er
e
ar
e
s
o
m
an
y
v
ar
iatio
n
s
in
tex
t
u
r
e
a
n
al
y
s
i
s
tec
h
n
iq
u
e
s
b
ec
au
s
e
o
f
th
e
tex
tu
r
es
h
a
v
e
d
if
f
er
en
t
t
y
p
es
o
f
p
atter
n
s
a
n
d
s
h
ap
es
o
n
t
h
e
s
u
r
f
ac
e.
Fo
r
ac
h
iev
in
g
b
etter
p
er
f
o
r
m
a
n
ce
,
d
i
f
f
er
e
n
t
t
y
p
es
o
f
f
e
atu
r
es
ar
e
e
x
tr
ac
ted
to
ch
ar
ac
ter
ize
th
e
tex
tu
r
e
i
m
ag
es.
T
ex
tu
r
e
an
al
y
s
i
s
an
d
in
v
esti
g
atio
n
s
ar
e
s
ig
n
i
f
ica
n
tl
y
ac
co
m
p
li
s
h
ed
in
o
n
e
o
f
t
h
e
t
w
o
w
a
y
s
,
i.e
.
s
tr
u
ctu
r
al
ap
p
r
o
ac
h
an
d
s
tati
s
tical
s
tr
ateg
y
.
Str
u
ct
u
r
al
ap
p
r
o
ac
h
es
s
en
tia
ll
y
f
o
cu
s
e
s
o
n
th
e
s
to
ch
a
s
tic
t
h
in
g
s
o
f
th
e
s
p
atial
cir
cu
latio
n
o
f
d
ar
k
lev
els
in
a
p
ictu
r
e.
I
n
f
in
d
i
n
g
th
e
c
h
ar
ac
ter
is
tic
s
o
f
an
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
A
N
o
ve
l A
p
p
r
o
a
ch
B
a
s
ed
o
n
Dec
r
ea
s
ed
Dimen
s
io
n
a
n
d
R
e
d
u
ce
d
Gra
y…
(
G
S
N
Mu
r
th
y
)
2503
i
m
a
g
e,
co
-
o
cc
u
r
r
en
ce
m
atr
ix
i
s
w
id
el
y
u
s
ed
.
Fro
m
th
e
co
-
e
v
en
t
l
a
ttice
s
et
o
f
te
x
t
u
r
al
ele
m
en
ts
s
ep
ar
ated
an
d
th
ese
co
m
p
o
n
e
n
ts
ar
e
b
r
o
ad
l
y
u
s
ed
to
r
e
m
o
v
e
te
x
t
u
r
al
d
ata
f
r
o
m
ad
v
an
ce
d
p
ict
u
r
es
[
4
]
[
5
]
.
I
n
b
asic
ap
p
r
o
ac
h
,
s
u
r
f
ac
e
is
co
n
s
id
er
ed
as
r
eiter
atio
n
o
f
a
f
e
w
p
r
i
m
iti
v
es.
Fo
r
s
u
r
f
ac
e
g
r
o
u
p
i
n
g
a
n
d
p
o
r
tr
ay
al,
th
ese
s
t
r
ate
g
ie
s
h
av
e
b
ee
n
co
n
n
ec
ted
b
y
a
f
e
w
cr
ea
to
r
s
an
d
m
ad
e
p
r
o
g
r
ess
to
a
s
p
ec
if
ic
d
eg
r
ee
[
6
]
.
So
m
an
y
ap
p
r
o
ac
h
es
ar
e
av
ail
ab
le
in
th
e
liter
at
u
r
e
f
o
r
tex
t
u
r
e
class
if
icatio
n
.
T
h
e
f
ir
s
t
an
d
to
p
m
o
s
t
ap
p
r
o
ac
h
is
L
o
ca
l
B
i
n
ar
y
P
att
er
n
(
L
B
P
)
ap
p
r
o
ac
h
[
7
]
[
8
]
.
B
u
t
L
B
P
ap
p
r
o
ac
h
h
as
s
o
m
e
d
i
s
ad
v
an
tag
e
s
.
I
f
th
e
ce
n
tr
al
p
ix
el
v
al
u
e
ch
a
n
g
es
b
y
1
,
th
e
L
B
P
v
alu
e
d
r
asti
ca
ll
y
ch
an
g
es.
Ot
h
er
ex
is
ti
n
g
ap
p
r
o
ac
h
es
ar
e
b
ased
o
n
w
a
v
elet
tr
a
n
s
f
o
r
m
[
9
]
[
1
0
]
,
s
tatis
tical
lear
n
in
g
f
r
o
m
m
o
r
p
h
o
l
o
g
ical
i
m
ag
e
p
r
o
ce
s
s
i
n
g
[
1
1
]
,
lo
n
g
l
in
ea
r
p
atte
r
n
s
[
1
2
]
[
1
3
]
,
e
d
g
e
d
ir
ec
tio
n
m
o
v
e
m
en
ts
[
1
4
]
,
ex
clu
d
i
n
g
C
o
m
p
lex
P
atter
n
s
[
1
5
]
an
d
p
r
ep
r
o
c
ess
ed
i
m
ag
e
s
[
1
6
]
.
T
ex
tu
r
e
p
ictu
r
es
ar
e
c
h
ar
ac
ter
ized
b
y
u
tili
z
in
g
d
i
f
f
er
en
t
w
a
v
elet
tr
an
s
f
o
r
m
s
u
s
in
g
s
tati
s
ti
ca
l
p
ar
a
m
eter
s
[
1
7
]
an
d
p
r
i
m
iti
v
e
p
ar
a
m
eter
s
.
R
ec
en
t
l
y
,
J
u
an
W
an
g
et.
al
[
1
8
]
p
r
o
p
o
s
ed
a
m
et
h
o
d
f
o
r
tex
t
u
r
e
clas
s
i
f
icatio
n
u
s
i
n
g
Scatter
in
g
Statis
t
ical
a
n
d
C
o
-
o
cc
u
r
r
en
ce
Featu
r
e
s
.
W
an
g
d
ev
elo
p
ed
n
e
w
ap
p
r
o
ac
h
f
o
r
tex
t
u
r
e
f
ea
t
u
r
es
ex
tr
ac
tio
n
.
T
h
is
ap
p
r
o
ac
h
u
s
ed
s
ca
tter
i
n
g
tr
an
s
f
o
r
m
f
o
r
s
ca
tter
in
g
s
tati
s
tica
l
f
ea
t
u
r
e
s
a
n
d
s
ca
t
ter
in
g
co
-
o
cc
u
r
r
en
ce
f
ea
t
u
r
es
ex
tr
ac
tio
n
w
h
ic
h
ar
e
d
er
iv
ed
f
r
o
m
s
u
b
-
b
an
d
s
o
f
t
h
e
s
ca
t
ter
i
n
g
d
ec
o
m
p
o
s
itio
n
a
n
d
o
r
ig
i
n
a
l
i
m
a
g
es
an
d
t
h
ese
f
ea
t
u
r
es a
r
e
u
s
ed
f
o
r
class
i
f
ica
tio
n
.
T
h
is
ap
p
r
o
ac
h
g
o
t r
ea
s
o
n
ab
le
p
er
ce
n
tag
e
r
ate
o
f
clas
s
i
f
i
ca
tio
n
b
u
t
th
e
t
i
m
e
co
m
p
le
x
it
y
is
m
o
r
e.
Siv
a
Ku
m
ar
et.
al
[
1
9
]
p
r
o
p
o
s
ed
a
m
et
h
o
d
f
o
r
s
to
n
e
tex
tu
r
e
class
if
ica
tio
n
b
a
s
ed
o
n
ed
g
e
d
ir
ec
tio
n
m
o
v
e
m
e
n
t.
I
n
th
is
ap
p
r
o
ac
h
,
ed
g
e
m
o
v
e
m
e
n
ts
ar
e
id
en
ti
f
ie
d
o
n
ea
ch
3
×3
s
u
b
-
i
m
ag
e
a
n
d
b
ased
o
n
th
e
ed
g
e
d
ir
ec
tio
n
m
o
v
e
m
e
n
t
s
,
t
h
e
te
x
t
u
r
e
i
m
ag
e
s
ar
e
clas
s
i
f
ied
.
T
h
is
ap
p
r
o
ac
h
m
ain
l
y
clas
s
if
ies t
h
e
tex
t
u
r
e
i
m
ag
e
in
to
t
w
o
g
r
o
u
p
s
o
n
l
y
a
n
d
ea
ch
g
r
o
u
p
co
n
s
is
ts
o
f
4
d
i
f
f
er
e
n
t
t
y
p
es
o
f
tex
t
u
r
e
i
m
a
g
es.
R
a
tn
a
B
h
ar
g
av
i
e
t
al
[
2
0
]
p
r
o
p
o
s
ed
an
ap
p
r
o
ac
h
f
o
r
d
etec
tio
n
o
f
L
es
io
n
u
s
in
g
tex
t
u
r
e
f
ea
tu
r
es
a
n
d
X
iao
r
o
n
g
Xu
e
et.
a
l
[
2
1
]
p
r
o
p
o
s
ed
an
ap
p
r
o
ac
h
f
o
r
C
lass
if
icatio
n
o
f
Fu
ll
y
P
o
lar
i
m
etr
ic
S
AR
I
m
a
g
es
b
ased
o
n
P
o
lar
i
m
etr
ic
Featu
r
es
an
d
Sp
atial
Featu
r
e
s
.
Vij
ay
Ku
m
ar
et.
al
[
2
2
]
p
r
o
p
o
s
ed
a
m
et
h
o
d
f
o
r
class
i
f
y
i
n
g
t
h
e
s
to
n
e
tex
t
u
r
es
in
to
f
o
u
r
ca
teg
o
r
ies
b
ased
o
n
o
cc
u
r
r
en
ce
o
f
T
-
p
atter
n
co
u
n
t
w
h
ic
h
ar
e
o
v
er
lap
p
ed
5
b
it
T
-
p
atter
n
s
o
n
ea
ch
5
×5
s
u
b
-
i
m
a
g
e.
T
h
e
class
i
f
icatio
n
r
ate
o
f
t
h
is
a
p
p
r
o
ac
h
is
ab
o
u
t
9
6
.
1
6
%
.
I
n
Vij
a
y
K
u
m
ar
’
s
w
o
r
k
,
s
ta
n
d
ar
d
class
i
f
icatio
n
alg
o
r
ith
m
s
ar
e
n
o
t u
s
ed
f
o
r
cla
s
s
i
f
y
in
g
t
h
e
s
to
n
e
te
x
tu
r
e
g
r
o
u
p
.
Stan
d
ar
d
class
i
f
icatio
n
s
y
s
t
e
m
s
co
n
s
u
m
e
m
o
r
e
ti
m
e
f
o
r
ex
tr
ac
tio
n
o
f
t
h
e
f
ea
tu
r
es f
r
o
m
s
to
n
e
i
m
ag
e
a
n
d
also
f
o
r
class
i
f
icatio
n
.
T
h
e
ex
is
ti
n
g
s
tan
d
ar
d
clas
s
i
f
i
ca
tio
n
ap
p
r
o
ac
h
es,
b
o
th
cla
s
s
i
f
icatio
n
o
f
s
to
n
e
te
x
tu
r
e
s
a
n
d
ex
tr
ac
tio
n
o
f
th
e
f
ea
tu
r
es
f
r
o
m
s
to
n
e
i
m
a
g
e
co
n
s
u
m
e
m
o
r
e
ti
m
e.
Oth
er
ex
is
ti
n
g
ap
p
r
o
ac
h
es
in
liter
at
u
r
e,
ev
en
p
r
o
p
o
s
ed
alg
o
r
ith
m
s
f
o
r
class
i
f
y
in
g
t
h
e
s
to
n
e
tex
tu
r
e
g
r
o
u
p
.
T
h
eir
class
if
ica
tio
n
r
es
u
lts
ar
e
n
o
t
co
m
p
ar
ed
w
i
th
s
tan
d
ar
d
class
i
f
icatio
n
al
g
o
r
ith
m
s
to
v
e
r
if
y
th
e
ac
c
u
r
ac
y
.
I
f
co
r
r
ec
t f
e
atu
r
es a
r
e
e
x
tr
ac
ted
th
e
n
t
h
e
y
f
it
f
o
r
b
o
th
s
ta
n
d
ar
d
class
i
f
icatio
n
an
d
also
f
o
r
u
s
e
r
d
ef
in
ed
alg
o
r
ith
m
.
So
,
th
e
p
r
esen
t
w
o
r
k
co
n
ce
n
tr
ate
s
o
n
d
ev
elo
p
in
g
a
m
et
h
o
d
ca
lled
DDRG
R
M
f
o
r
class
if
y
i
n
g
t
h
e
s
to
n
e
tex
tu
r
es i
n
to
f
o
u
r
g
r
o
u
p
s
.
T
ill
n
o
w
m
aj
o
r
ity
o
f
t
h
e
ex
is
tin
g
tech
n
iq
u
es
e
x
tr
ac
t
f
ea
tu
r
es
f
r
o
m
th
e
e
n
tire
i
m
a
g
e.
T
h
e
p
r
o
p
o
s
ed
DDRG
R
M
s
tr
ate
g
y
is
to
d
ec
r
ea
s
e
th
e
s
to
n
e
i
m
ag
e
d
i
m
e
n
s
io
n
alit
y
i
n
to
(
2
N/5
×2
M/5
)
an
d
ap
p
lies
f
u
zz
y
co
n
ce
p
t
f
o
r
les
s
en
i
n
g
t
h
e
d
i
m
le
v
el
r
a
n
g
e
f
o
r
v
iab
le
a
n
d
p
r
o
f
icien
t
s
to
n
e
s
u
r
f
ac
e
g
r
o
u
p
in
g
.
An
o
th
er
f
u
n
d
a
m
en
ta
l
is
s
u
e
in
clas
s
i
f
i
ca
tio
n
o
f
tex
t
u
r
e
a
n
d
r
ec
o
g
n
itio
n
is
tex
t
u
r
e
c
h
ar
ac
ter
iza
tio
n
f
r
o
m
d
er
iv
ed
f
ea
t
u
r
es.
Ma
n
y
o
f
th
e
ex
i
s
ti
n
g
ap
p
r
o
ac
h
es
h
av
e
t
h
e
d
r
a
w
b
ac
k
o
f
co
m
p
u
tatio
n
al
co
m
p
le
x
it
y
as
t
h
e
y
in
c
lu
d
e
p
r
o
ce
s
s
in
g
o
f
en
t
ir
e
i
m
a
g
e
w
it
h
lar
g
e
r
an
g
e
o
f
g
r
a
y
le
v
els
f
o
r
tex
t
u
r
e
class
if
ica
tio
n
an
d
r
ec
o
g
n
itio
n
.
T
o
ad
d
r
ess
th
is
,
th
e
p
r
esen
t
p
ap
er
p
r
o
p
o
s
es
an
ap
p
r
o
ac
h
in
w
h
ic
h
th
e
i
m
ag
e
d
i
m
en
s
io
n
a
n
d
d
i
m
lev
el
r
an
g
e
ar
e
d
ec
r
ea
s
ed
w
it
h
n
o
lo
s
s
o
f
s
u
r
f
ac
e
co
m
p
o
n
e
n
t d
ata.
T
h
e
m
ai
n
o
b
j
ec
tiv
e
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
to
b
e
co
m
p
atib
le
w
it
h
b
o
th
th
e
ap
p
r
o
ac
h
es
i.e
.
f
o
r
u
s
er
d
e
f
i
n
ed
al
g
o
r
ith
m
a
n
d
al
s
o
f
o
r
s
ta
n
d
ar
d
clas
s
i
f
icatio
n
al
g
o
r
ith
m
s
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
d
o
es n
o
t
u
s
e
an
y
s
tan
d
ar
d
clas
s
i
f
icatio
n
alg
o
r
it
h
m
s
f
o
r
clas
s
i
f
y
i
n
g
t
h
e
s
to
n
e
t
ex
tu
r
e
g
r
o
u
p
.
T
h
e
r
est
o
f
th
e
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
.
Sec
tio
n
2
d
escr
ib
es
t
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
.
D
er
iv
ed
u
s
er
d
ef
in
ed
al
g
o
r
ith
m
a
n
d
R
e
s
u
lt
s
ar
e
e
x
p
lain
e
d
in
s
ec
tio
n
3
.
Fin
all
y
,
co
n
cl
u
s
io
n
s
ar
e
g
i
v
e
n
in
s
ec
tio
n
4
.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
Fo
r
p
o
r
tr
ay
in
g
th
e
at
tr
ib
u
tes
o
f
th
e
n
e
ig
h
b
o
r
h
o
o
d
ex
a
m
p
le
o
f
th
e
s
u
r
f
ac
e
b
y
u
til
izi
n
g
s
u
r
f
ac
e
d
escr
ip
to
r
s
tr
ateg
ies,
f
o
r
ex
a
m
p
le,
L
o
ca
l
B
i
n
ar
y
P
atter
n
(
L
B
P
)
,
T
ex
tu
r
e
U
n
it
(
T
U)
an
d
T
ex
to
n
s
.
T
h
e
s
u
r
f
ac
e
d
escr
ip
to
r
s
ar
e
v
alu
ab
le
f
o
r
s
u
r
f
ac
e
ex
a
m
i
n
atio
n
an
d
cr
itical
g
r
o
u
p
in
g
a
n
d
it
g
iv
e
s
b
o
th
f
a
ctu
al
a
n
d
au
x
iliar
y
q
u
alities
o
f
a
s
u
r
f
ac
e.
T
h
ese
d
escr
ip
to
r
s
ar
e
to
tally
n
ea
r
b
y
an
d
g
en
er
all
y
ch
ar
ac
t
er
ized
o
n
a
3
×
3
n
eig
h
b
o
r
h
o
o
d
.
T
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e
d
is
p
la
y
tak
e
s
a
5
×5
n
eig
h
b
o
r
h
o
o
d
,
an
d
r
ed
u
ce
s
it
in
to
a
2
×2
n
eig
h
b
o
r
h
o
o
d
w
ith
o
u
t
lo
s
s
o
f
a
n
y
s
u
r
f
a
ce
d
ata
an
d
f
u
r
th
er
it
d
i
m
i
n
i
s
h
e
s
t
h
e
d
i
m
le
v
el
r
an
g
e
u
tili
zi
n
g
f
l
u
f
f
y
r
atio
n
ale.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
5
,
Octo
b
er
2
0
1
7
:
2
5
0
2
–
2
5
1
3
2504
T
h
e
p
r
o
p
o
s
ed
DDRGR
M
m
o
d
el
m
ain
l
y
co
n
s
is
t
s
o
f
6
s
tep
s
.
I
n
s
tep
1
,
co
n
v
er
t
th
e
R
GB
s
to
n
e
tex
t
u
r
e
i
m
a
g
e
i
n
to
Gr
a
y
le
v
el
i
m
a
g
e
u
s
i
n
g
W
eig
h
ted
R
GB
co
n
v
er
s
io
n
m
et
h
o
d
.
Fo
r
m
atio
n
o
f
n
i
n
e
o
v
er
lap
p
ed
s
u
b
3
×3
s
u
b
i
m
ag
e
s
f
r
o
m
a
5
×5
s
u
b
i
m
a
g
e
is
p
er
f
o
r
m
ed
i
n
s
t
ep
2
.
I
n
s
tep
3
Der
iv
a
tio
n
o
f
“L
o
ca
l
D
if
f
er
en
ce
Ma
tr
ix
(
L
DM
)
”
o
n
t
h
e
n
in
e
o
v
er
lap
p
ed
3
×3
s
u
b
i
m
ag
e
s
a
n
d
g
en
er
ate
t
h
e
r
ed
u
ce
d
m
a
tr
ix
.
Fu
r
th
er
r
ed
u
ce
th
e
3
×3
s
u
b
im
a
g
e
in
to
2
×2
s
u
b
im
a
g
e
w
i
th
o
u
t
lo
s
in
g
th
e
te
x
t
u
r
e
im
a
g
e
in
f
o
r
m
at
io
n
in
s
tep
4
.
Step
5
,
r
ed
u
ce
th
e
g
r
a
y
le
v
el
r
an
g
e
i
n
ea
ch
2
×2
s
u
b
i
m
ag
e
u
s
in
g
f
u
zz
y
co
n
ce
p
t
an
d
g
en
er
ate
t
h
e
F
u
zz
y
r
ed
u
ce
d
co
-
o
cc
u
r
r
en
ce
m
atr
i
x
,
in
s
tep
6
,
ex
tr
ac
t
th
e
C
M
f
ea
tu
r
e
s
f
o
r
class
i
f
icatio
n
.
T
h
e
b
lo
ck
d
iag
r
am
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
s
h
o
w
n
in
F
ig
u
r
e
1
.
Us
e
T
h
re
sh
o
ld
in
g
Fig
u
r
e
1
.
B
lo
ck
d
iag
r
a
m
o
f
D
DR
G
R
M
Mo
d
el
2
.
1
.
Co
nv
er
t
RG
B
t
o
G
ra
y
le
v
el
i
m
a
g
e
:
T
o
ex
tr
ac
t
th
e
f
ea
tu
r
e
s
th
e
R
GB
i
m
ag
e
w
i
ll
b
e
tr
an
s
f
o
r
m
e
d
to
Gr
ay
i
m
ag
e
u
s
in
g
W
eig
h
ted
R
GB
co
n
v
er
s
io
n
.
As
th
e
R
GB
i
m
a
g
e
is
f
o
r
m
ed
b
y
3
co
m
m
a
n
d
ed
h
u
e
s
i.e
.
R
ed
,
Gr
ee
n
an
d
B
lu
e,
in
W
eig
h
ted
R
GB
co
n
v
er
s
io
n
d
iv
er
s
e
w
eig
h
t
s
ar
e
ass
ig
n
ed
to
ea
ch
s
h
ad
in
g
s
eg
m
e
n
t
an
d
th
ese
t
h
r
ee
s
eg
m
en
ts
ar
e
u
s
ed
f
o
r
co
n
v
er
ti
n
g
t
h
e
R
GB
i
m
a
g
e
to
g
r
a
y
i
m
a
g
e.
T
h
e
tr
an
s
f
o
r
m
atio
n
p
r
o
c
ed
u
r
e
is
s
p
ec
if
ied
in
eq
u
atio
n
1
.
(
,
)
=
0
.
3
∗
(
,
)
+
0
.
59
∗
(
,
)
+
0
.
11
∗
(
,
)
(
1
)
W
h
er
e
R
,
G,
B
ar
e
th
e
R
ed
,
Gr
ee
n
,
B
lu
e
co
lo
r
co
m
p
o
n
en
t
v
al
u
es,
(
x
,
y
)
ar
e
th
e
p
i
x
el
p
o
s
itio
n
s
an
d
Gr
a
y
(
x
,
y
)
r
ep
r
ese
n
ts
t
h
e
g
r
a
y
v
al
u
e
at
t
h
e
g
i
v
e
n
p
ix
el
p
o
s
itio
n
(
x
,
y
)
.
T
h
e
R
GB
i
m
a
g
e
an
d
r
esu
l
tan
t
g
r
a
y
i
m
a
g
e
af
ter
co
n
v
er
s
io
n
ar
e
s
h
o
w
n
i
n
F
i
g
u
r
e
2
.
(
a)
(
b
)
Fig
u
r
e
2
.
Ma
r
b
le
s
to
n
e
i
m
a
g
e
(
a)
C
o
lo
r
im
a
g
e
(
b
)
R
esu
lta
n
t
Gr
a
y
lev
el
i
m
a
g
e
2
.
2
.
F
o
r
m
a
t
io
n o
f
9
o
v
er
la
pp
ed
3
×
3
s
ub
i
m
a
g
es f
ro
m
a
5
×
5
s
ub
i
m
a
g
e
:
T
h
e
5
×5
s
u
b
im
a
g
e
co
n
s
is
t
s
o
f
2
5
p
ix
els
r
ep
r
esen
ted
b
y
{
V1
,
V2
,
….
,
V1
3
,
.
.
.
V2
5
},
w
h
er
e
V1
3
r
ep
r
esen
ts
t
h
e
g
r
a
y
v
alu
e
o
f
th
e
i
n
n
er
m
o
s
t
ce
n
ter
p
ix
el
a
n
d
r
e
m
ain
in
g
ar
e
th
e
n
ei
g
h
b
o
r
in
g
p
ix
e
l
i
n
te
n
s
it
y
v
alu
e
s
as
s
h
o
w
n
i
n
Fig
u
r
e
3
.
Fig
u
r
e
4
r
ep
r
esen
ts
o
v
er
l
ap
p
ed
3
×3
s
u
b
w
in
d
o
w
s
r
e
f
er
r
ed
as
{
w
1
,
w
2
,
w
3
,
…
w
9
}
ex
tr
ac
ted
f
r
o
m
t
h
e
5
X5
s
u
b
i
m
a
g
e
r
ep
r
esen
ted
in
F
ig
u
r
e
3
.
C
o
n
v
e
r
t
R
G
B
t
o
G
r
a
y
l
e
v
e
l
i
mag
e
F
o
r
mat
i
o
n
o
f
9
3
×
3
su
b
i
m
a
g
e
s
Ex
t
r
a
c
t
3
C
M
F
e
a
t
u
r
e
s
D
e
r
i
v
e
t
h
e
U
se
r
d
e
f
i
n
e
d
A
l
g
o
r
i
t
h
m
Est
i
m
a
t
e
S
t
o
n
e
t
e
x
t
u
r
e
G
r
o
u
p
G
e
n
e
r
a
t
e
L
D
M
D
e
c
r
e
a
se
t
h
e
3
×
3
s
u
b
i
mag
e
i
n
t
o
2
×
2
s
u
b
i
mag
e
R
e
d
u
c
e
t
h
e
g
r
a
y
l
e
v
e
l
r
a
n
g
e
I
n
p
u
t
S
t
o
n
e
i
mag
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
A
N
o
ve
l A
p
p
r
o
a
ch
B
a
s
ed
o
n
Dec
r
ea
s
ed
Dimen
s
io
n
a
n
d
R
e
d
u
ce
d
Gra
y…
(
G
S
N
Mu
r
th
y
)
2505
V
1
V
2
V
3
V
4
V
5
V
6
V
7
V
8
V
9
V
10
V
11
V
12
V
13
V
14
V
15
V
16
V
17
V
18
V
19
V
20
V
21
V
22
V
23
V
24
V
25
Fig
u
r
e
3.
R
ep
r
esen
tatio
n
o
f
a
5
×5
s
u
b
i
m
ag
e
V
1
V
2
V
3
V
2
V
3
V
4
V
3
V
4
V
5
V
6
V
7
V
8
V
7
V
8
V
9
V
8
V
9
V
10
V
11
V
12
V
13
V
12
V
13
V
14
V
13
V
14
V
15
w
1
w
2
w
3
V
6
V
7
V
8
V
7
V
8
V
9
V
8
V
9
V
10
V
11
V
12
V
13
V
12
V
13
V
14
V
13
V
14
V
15
V
16
V
17
V
18
V
17
V
18
V
19
V
18
V
19
V
20
w
4
w
5
w
6
V
11
V
1
2
V
13
V
12
V
13
V
14
V
13
V
14
V
15
V
16
V
17
V
18
V
17
V
18
V
19
V
18
V
19
V
20
V
21
V
22
V
23
V
22
V
23
V
24
V
23
V
24
V
25
w
7
w
8
w
9
Fig
u
r
e
4.
Fo
r
m
a
tio
n
o
f
o
v
er
la
p
p
ed
3
×
3
n
eig
h
b
o
r
h
o
o
d
s
{
w
1
,
w
2
,
w
3
,
…,
w
9
}
f
r
o
m
F
ig
u
r
e
3
2
.
3
.
Der
iv
a
t
io
n o
f
L
D
M
o
n e
a
ch
3
×
3
ove
rla
pp
ed
w
ind
o
w
o
f
5
×
5
s
ub
i
m
a
g
e
:
I
n
th
i
s
s
tep
,
L
DM
is
f
i
g
u
r
ed
f
o
r
ev
er
y
o
n
e
o
f
t
h
e
n
i
n
e
3
×3
co
v
er
ed
w
in
d
o
w
s
{
w
1
,
w
2
,
w3
,
…
,
w
9
}
o
f
5
×5
s
u
b
p
ictu
r
e.
T
h
e
L
D
M
g
iv
e
s
a
p
r
o
d
u
ctiv
e
p
o
r
tr
ay
al
o
f
s
u
r
f
ac
e
p
ictu
r
e.
T
h
e
L
DM
o
n
ea
ch
w
i
i
s
th
e
o
u
tr
ig
h
t
co
n
tr
a
s
t
b
et
w
ee
n
th
e
n
eig
h
b
o
r
in
g
p
ix
el
an
d
t
h
e
d
ar
k
esti
m
at
io
n
o
f
th
e
f
o
ca
l
p
ix
e
l
w
h
ic
h
is
e
v
al
u
ated
u
s
i
n
g
eq
u
a
tio
n
2
an
d
r
ep
r
esen
ted
i
n
F
ig
u
r
e
5
.
T
h
i
s
r
es
u
lt
s
in
n
in
e
n
e
w
3
×3
L
DM
s
r
ep
r
esen
ted
a
s
{
L
DM
1
,
L
DM
2
,
L
DM
3
,
…,
L
DM
9
}
f
o
r
ea
ch
o
v
er
lap
p
ed
w
i
n
d
o
w
{
w
1,
w
2
….
w
9
}
.
L
DM
i
=
ab
s
(
v
i
-
v
c
)
f
o
r
i =
1
,
2
,
.
.
.
9
(
2
)
W
h
er
e
v
c
is
th
e
ce
n
tr
e
p
ix
el
an
d
v
i
r
ep
r
esen
t
t
h
e
n
ei
g
h
b
o
r
in
g
p
i
x
el
v
al
u
e
s
o
f
th
e
o
v
er
l
ap
p
ed
3
×3
n
eig
h
b
o
r
h
o
o
d
.
B
asin
g
o
n
eq
u
atio
n
2
th
e
r
es
u
ltan
t
v
al
u
e
o
f
ea
ch
L
D
M
in
w
h
ic
h
th
e
ce
n
tr
al
p
ix
el
v
al
u
e
i
s
al
w
a
y
s
ze
r
o
.
│V
1
-
V
7
│
│V
2
-
V
7
│
│V
3
-
V
7
│
│V
6
-
V
7
│
│V
7
-
V
7
│
│V
8
-
V
7
│
│V
11
-
V
7
│
│V
12
-
V
7
│
│V
13
-
V
7
│
Fig
u
r
e
5
.
Gen
er
atio
n
o
f
L
DM
1
f
r
o
m
w
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
5
,
Octo
b
er
2
0
1
7
:
2
5
0
2
–
2
5
1
3
2506
2
.
4
.
G
ener
a
t
io
n o
f
Dec
re
a
s
e
Di
m
en
s
io
n M
a
t
rix
(
DD
M
)
o
f
5
×
5
into
3
×
3
w
ind
o
w
:
I
n
th
i
s
ea
ch
v
al
u
e
o
f
D
DM
is
ev
alu
a
ted
f
r
o
m
ea
c
h
o
f
t
h
e
n
i
n
e
L
DM
’
s
g
en
er
ated
i
n
t
h
e
p
r
ev
io
u
s
s
tep
in
t
w
o
s
ta
g
es
:
g
e
n
er
atio
n
o
f
Me
an
L
DM
i
n
th
e
f
ir
s
t
s
tep
a
n
d
th
e
n
g
e
n
er
ate
DDM
.
I
n
s
ta
g
e
o
n
e,
t
h
e
m
ea
n
o
f
th
e
9
w
i
n
d
o
w
s
w
h
ic
h
ar
e
g
en
er
ated
in
p
r
ev
io
u
s
s
tep
b
y
u
s
i
n
g
t
h
e
eq
u
atio
n
3
ar
e
f
o
u
n
d
.
T
h
e
g
e
n
er
ated
v
alu
e
s
f
o
r
m
s
a
m
atr
ix
is
ca
lled
Me
an
L
DM
(
ML
DM
)
.
T
h
e
ML
D
M
is
a
3
×3
w
i
n
d
o
w
w
it
h
n
in
e
el
e
m
en
ts
(
M
L
DP
1
to
ML
DP
9
).
T
h
e
ML
DM
p
r
eser
v
es th
e
lo
ca
l r
eg
io
n
p
o
s
s
es
s
io
n
s
in
clu
d
i
n
g
ed
g
e
in
f
o
r
m
a
tio
n
.
ML
DP
i
=
m
ea
n
o
f
(
L
DM
i
)
f
o
r
i =
1
,
2
,
…9
(
3
)
Fu
r
t
h
er
,
g
en
er
ate
t
h
e
DDM
b
y
ca
lcu
lati
n
g
t
h
e
lo
ca
l
d
if
f
er
en
ce
b
et
w
ee
n
t
h
e
n
eig
h
b
o
r
in
g
p
ix
el
v
al
u
e
s
an
d
ce
n
tr
al
p
ix
el
v
al
u
e
o
f
t
h
e
ML
DP
m
atr
i
x
an
d
i
s
r
ep
r
esen
t
ed
b
y
eq
u
atio
n
4
.
DDM
P
i
=
ab
s
(
ML
DP
i
–
ML
DP
c
)
f
o
r
ML
DP
i
=
1
,
2
,
…9
(
4
)
T
h
e
E
q
u
atio
n
4
r
ev
ea
ls
th
at
co
n
ti
n
u
o
u
s
l
y
d
o
m
i
n
an
t p
ix
el
v
al
u
e
o
f
t
h
e
3
×3
DDM
is
ze
r
o
.
2
.
5
.
G
ener
a
t
io
n o
f
Reduced
Di
m
en
s
io
n M
a
t
rix
(
RD
M
)
o
f
2
×
2
w
ind
o
w
f
ro
m
DDM
:
T
h
e
g
en
er
atio
n
p
r
o
ce
s
s
o
f
R
DM
m
ar
i
x
i
s
s
h
o
w
n
in
f
i
g
u
r
e
6
.
T
h
e
DDM
w
i
n
d
o
w
co
m
p
r
is
es
o
f
n
in
e
q
u
alities
w
h
ich
is
cr
ea
ted
i
n
p
r
ev
io
u
s
s
tep
as
s
h
o
w
n
i
n
f
i
g
u
r
e
6
(
a)
.
I
n
th
i
s
p
r
o
g
r
es
s
io
n
,
th
e
DD
M
o
f
a
3
×3
n
eig
h
b
o
r
h
o
o
d
is
les
s
en
ed
i
n
to
a
2
×2
R
DM
b
y
u
tili
zi
n
g
T
r
ian
g
u
lar
Sh
ap
e
P
r
i
m
it
iv
e
s
(
T
SP
)
.
T
h
e
p
r
o
p
o
s
ed
T
SP
is
an
ass
o
ciate
d
n
ei
g
h
b
o
r
h
o
o
d
o
f
th
r
ee
p
i
x
els
o
n
a
3
×3
DDM
,
w
it
h
o
u
t
f
o
ca
l
p
ix
e
l.
T
h
e
T
SP
'
s
o
n
DD
M
do
esn
’
t
co
n
s
id
er
f
o
ca
l
p
ix
el
as
it
s
d
ar
k
lev
el
is
co
n
s
tan
tl
y
ze
r
o
.
T
h
e
n
o
r
m
al
o
f
t
h
ese
T
SP
'
s
cr
ea
tes
p
i
x
el
esti
m
atio
n
s
o
f
R
ed
u
ce
d
Di
m
en
s
io
n
Ma
tr
i
x
(
R
DM
)
o
f
m
e
asu
r
e
2
×2
as
ap
p
ea
r
ed
in
Fi
g
u
r
e
6
(
b
)
b
ased
o
n
eq
u
atio
n
s
5
to
8
.
B
y
t
h
is
t
h
e
p
r
o
p
o
s
ed
tech
n
iq
u
e
d
ec
r
e
ase
s
th
e
te
x
t
u
r
e
i
m
a
g
e
o
f
s
ize
N×
M
in
to
th
e
s
ize
(
2
N/5
)
×
(
2
M/5
)
.
R
DM
P
1
=
(
DDM
P
1
+
DDM
P
2
+D
DM
P
4
)
/ 3
(
5
)
R
DM
P
2
=
(
DDM
P
2
+
DDM
P
3
+D
DM
P
6
)
/ 3
(
6
)
R
DM
P
3
=
(
DDM
P
4
+
DDM
P
7
+D
DM
P
8
)
/ 3
(
7
)
R
DM
P
4
=
(
DDM
P
6
+
DDM
P
8
+D
DM
P
9
)
/ 3
(
8
)
DD
M
P
1
DD
M
P
2
DD
M
P
3
DD
M
P
4
DD
M
P
5
DD
M
P
6
RDMP
1
RDMP
2
DD
M
P
7
DD
M
P
8
DD
M
P
9
RDMP
3
RDMP
4
(
a)
(
b
)
Fig
u
r
e
6.
Gen
er
atio
n
p
r
o
ce
s
s
o
f
a
R
DM
o
f
s
ize
2
×2
f
r
o
m
a
3
×3
DDM
n
eig
h
b
o
r
h
o
o
d
.
a)
T
h
e
DDM
n
ei
g
h
b
o
r
h
o
o
d
b
)
R
DM
.
2
.
6
.
Reduct
io
n o
f
g
ra
y
le
v
el
ra
ng
e
in RDM
us
ing
f
uzzy
lo
g
ic
:
Fu
zz
y
r
atio
n
ale
h
as
ce
r
tai
n
r
ea
l
f
o
ca
l
p
o
in
ts
o
v
er
co
n
v
e
n
t
io
n
al
B
o
o
lean
r
atio
n
ale
w
it
h
r
eg
ar
d
s
to
ce
r
tif
iab
le
ap
p
licatio
n
s
,
f
o
r
e
x
a
m
p
le,
s
u
r
f
ac
e
p
o
r
tr
a
y
al
o
f
g
e
n
u
i
n
e
p
ict
u
r
es.
T
o
d
ea
l p
r
ec
is
el
y
w
i
th
th
e
ar
ea
s
o
f
r
eg
u
lar
p
ictu
r
es
e
v
en
w
i
th
i
n
t
h
e
s
i
g
h
t
o
f
cla
m
o
r
an
d
th
e
d
iv
er
s
e
p
r
o
ce
d
u
r
es
o
f
s
u
b
titl
e
an
d
d
ig
itizatio
n
f
l
u
f
f
y
r
atio
n
ale
is
p
r
esen
ted
o
n
DD
M.
T
h
e
p
r
o
p
o
s
ed
f
lu
f
f
y
r
atio
n
ale
co
n
v
er
t
s
DD
M
d
ar
k
le
v
el
s
in
to
5
le
v
els
r
an
g
i
n
g
f
r
o
m
0
to
4
.
T
h
e
r
esu
ltan
t
f
r
a
m
e
w
o
r
k
is
ca
lled
Dec
r
ea
s
e
Di
m
e
n
s
io
n
R
ed
u
c
in
g
G
r
a
y
le
v
el
R
a
n
g
e
Ma
tr
i
x
(
DDRGR
M)
.
I
n
L
B
P
d
o
u
b
le
ex
a
m
p
le
s
ar
e
a
s
s
e
s
s
ed
b
y
co
n
tr
asti
n
g
t
h
e
n
ei
g
h
b
o
r
in
g
p
ix
els
a
n
d
f
o
ca
l
p
ix
el.
T
h
e
p
r
o
p
o
s
ed
DDRGR
M
m
o
d
el
is
d
eter
m
i
n
ed
b
y
lo
o
k
in
g
at
th
e
ev
er
y
p
i
x
el
o
f
th
e
2
×2
DDM
w
it
h
th
e
n
o
r
m
al
p
ix
el
esti
m
at
io
n
s
o
f
t
h
e
DD
M.
T
h
e
DDRGR
M
p
o
r
tr
a
y
al
is
ap
p
ea
r
ed
in
F
ig
u
r
e
7
.
T
h
e
ac
co
m
p
an
y
i
n
g
E
q
u
atio
n
s
9
is
u
tili
ze
d
to
d
ec
id
e
th
e
co
m
p
o
n
en
t
s
o
f
DD
R
G
R
M
m
o
d
el
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
A
N
o
ve
l A
p
p
r
o
a
ch
B
a
s
ed
o
n
Dec
r
ea
s
ed
Dimen
s
io
n
a
n
d
R
e
d
u
ce
d
Gra
y…
(
G
S
N
Mu
r
th
y
)
2507
Fig
u
r
e
7
.
Fu
zz
y
r
ep
r
ese
n
tatio
n
o
f
DDRG
R
M
m
o
d
el
o
f
t
h
e
i
m
ag
e
0
if
R
DM
P
i
< V
0
an
d
R
DM
P
i
<
x
1
if
R
DM
P
i
< V
0
an
d
R
DM
P
i
≥
x
DDRG
R
MP
i
=
2
if
R
DM
P
i
= V
0
f
o
r
i =
1
,
2
,
3
,
4
(
9
)
3
if
R
DM
P
i
> V
0
an
d
R
DM
P
i
>y
4
if
R
DM
P
i
> V
0
an
d
R
DM
P
i
≤
y
W
h
er
e
x
,
y
ar
e
th
e
u
s
er
-
s
p
ec
i
f
i
ed
v
alu
es a
n
d
V
0
=
(
∑
T
S
P
i
=
)
(
1
0
)
Fo
r
ex
a
m
p
le,
th
e
p
r
o
ce
s
s
o
f
ev
alu
ati
n
g
DDRG
R
M
m
o
d
e
l
f
r
o
m
a
s
u
b
R
DM
i
m
a
g
e
o
f
2
×2
is
s
h
o
w
n
in
F
ig
u
r
e
8
.
T
h
e
Fig
u
r
e
8
(
a)
r
ep
r
esen
t
s
R
DM
an
d
f
ig
u
r
e
8
(
b
)
r
ep
r
esen
ts
th
e
r
s
u
lten
t
f
u
zz
y
m
a
tr
ix
f
r
o
m
R
DM
.
I
n
th
i
s
s
t
u
d
y
,
x
a
n
d
y
v
alu
e
s
ar
e
ch
o
s
en
a
s
V
0
/2
an
d
3
V
0
/2
r
e
s
p
ec
tiv
el
y
.
28
39
1
2
61
9
4
0
(
a)
(
b
)
Fig
u
r
e
8
.
T
h
e
p
r
o
ce
s
s
o
f
ev
al
u
atin
g
DDR
GR
M
m
o
d
el
f
r
o
m
s
u
b
R
DM
(
a)
R
DM
(
b
)
DDRGR
M
m
o
d
el
2
.
7
.
Co
m
p
uta
t
io
n o
f
C
M
f
e
a
t
ures o
n t
he
deriv
ed
DDR
G
R
M
m
o
del
:
T
h
e
p
r
esen
t
ap
p
r
o
ac
h
d
eter
m
in
ed
Gr
a
y
L
ev
e
l
C
o
-
o
cc
u
r
r
en
ce
Ma
tr
ix
(
G
L
C
M)
o
n
t
h
e
DDRG
R
M
m
o
d
el
o
f
th
e
s
to
n
e
tex
tu
r
e
i
m
ag
e.
GL
C
M
is
p
r
o
p
o
s
ed
b
y
Har
alick
to
ch
ar
ac
ter
ize
th
e
i
m
ag
e
b
ased
o
n
h
o
w
ce
r
tain
d
ar
k
lev
els
h
ap
p
en
i
n
co
m
p
ar
is
o
n
w
it
h
o
th
er
d
im
le
v
els.
G
L
C
M
ca
n
g
au
g
e
th
e
s
u
r
f
ac
e
o
f
th
e
p
ictu
r
e
s
in
ce
co
-
ev
e
n
t
f
r
a
m
e
w
o
r
k
s
a
r
e
g
en
er
all
y
v
ast
a
n
d
s
ca
n
t
y
.
GL
C
M
is
co
n
s
id
er
ed
to
b
e
a
b
en
ch
m
ar
k
f
o
r
ex
tr
ac
ti
n
g
Har
alic
k
f
ea
t
u
r
es
li
k
e
an
g
u
lar
s
ec
o
n
d
m
o
m
e
n
t,
c
o
n
tr
ast,
co
r
r
elatio
n
,
v
ar
ia
n
ce
,
in
v
er
s
e
d
if
f
er
en
c
e
m
o
m
e
n
t,
s
u
m
a
v
er
ag
e,
s
u
m
v
ar
ian
ce
,
s
u
m
en
tr
o
p
y
,
e
n
tr
o
p
y
,
d
i
f
f
er
e
n
ce
v
ar
ian
ce
,
d
if
f
er
e
n
ce
e
n
tr
o
p
y
,
in
f
o
r
m
atio
n
m
ea
s
u
r
es
o
f
co
r
r
elatio
n
an
d
m
a
x
i
m
al
co
r
r
elat
io
n
co
ef
f
icien
t,
etc.
.
T
h
ese
ele
m
e
n
ts
h
av
e
b
ee
n
b
r
o
ad
ly
u
tili
ze
d
as
a
p
ar
t
o
f
th
e
in
v
es
tig
a
tio
n
,
g
r
o
u
p
i
n
g
a
n
d
elu
cid
atio
n
o
f
p
ictu
r
e
i
n
f
o
r
m
a
tio
n
.
I
ts
p
o
in
t
is
to
p
o
r
tr
ay
th
e
s
to
ch
a
s
tic
p
r
o
p
er
ties
o
f
t
h
e
s
p
atial
co
n
v
e
y
an
ce
o
f
d
ar
k
le
v
els
i
n
a
n
i
m
ag
e.
O
u
t
o
f
th
e
s
e
p
r
o
p
o
s
ed
Har
alick
f
ea
tu
r
es
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
u
s
ed
th
r
ee
Har
alick
h
ig
h
li
g
h
ts
i.e
.
C
o
r
r
elat
io
n
(
C
R
)
,
C
l
u
s
ter
P
r
o
m
in
e
n
ce
(
C
P
)
an
d
I
n
f
o
r
m
atio
n
m
ea
s
u
r
e
o
f
co
r
r
elatio
n
1
(
I
MC1
)
f
o
r
class
if
ica
tio
n
o
f
s
to
n
e
te
x
t
u
r
esi
n
to
4
d
if
f
er
e
n
t
g
r
o
u
p
s
.
Fo
r
ch
ar
ac
ter
izatio
n
o
f
s
to
n
e
tex
t
u
r
es
i
n
to
4
u
n
iq
u
e
g
r
o
u
p
s
eq
u
at
io
n
s
(
11
)
to
(
13
)
a
r
e
u
s
ed
.
T
h
e
DDRGR
M
m
et
h
o
d
w
it
h
GL
C
M
co
n
s
o
lid
ate
s
th
e
b
en
e
f
its
o
f
b
o
th
s
ta
tis
tica
l
an
d
s
tr
u
c
tu
r
al
in
f
o
r
m
a
tio
n
o
f
th
e
s
to
n
e
tex
t
u
r
e
i
m
ag
e.
=
∑
∑
(
∗
)
∗
(
,
)
−
(
∗
)
∗
=
1
=
1
(
1
1
)
W
h
er
e
,
an
d
,
ar
e
th
e
m
ea
n
a
n
d
s
tan
d
ar
d
d
ev
iatio
n
s
o
f
p
r
o
b
ab
i
lit
y
m
atr
i
x
GL
C
M
a
lo
n
g
r
o
w
w
i
s
e
x
an
d
co
lu
m
n
w
i
s
e
y
=
∑
∑
(
+
−
−
)
4
=
1
=
1
∗
(
,
)
(
1
2
)
1
=
∑
∑
(
∗
)
∗
(
(
,
)
∗
=
1
=
1
(
1
3
)
W
h
er
e
P
ij
is
th
e
p
ix
el
v
al
u
e
o
f
th
e
i
m
ag
e
at
p
o
s
itio
n
(
i,
j
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
5
,
Octo
b
er
2
0
1
7
:
2
5
0
2
–
2
5
1
3
2508
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
p
r
o
p
o
s
ed
DDRG
R
M
m
o
d
el
w
ith
C
M
f
ea
tu
r
e
s
i
s
i
m
p
l
e
m
en
ted
u
s
i
n
g
a
d
ata
s
et
o
f
6
1
2
s
to
n
e
i
m
a
g
es
co
llected
f
r
o
m
Ma
y
a
n
g
d
atab
ase,
6
7
8
s
to
n
e
i
m
a
g
es
co
llected
f
r
o
m
V
is
T
ex
d
atab
ase,
8
3
2
im
a
g
es
co
llected
f
r
o
m
P
au
l
B
o
u
r
k
e
d
atab
ase,
4
0
0
s
to
n
e
te
x
t
u
r
e
i
m
a
g
es
co
llec
ted
f
r
o
m
Go
o
g
le
d
ata
b
ase
i.e
.
,
a
to
tal
o
f
2
5
2
2
s
to
n
e
tex
tu
r
e
i
m
a
g
e
s
.
Sa
m
p
le
i
m
a
g
e
s
o
f
ea
ch
g
r
o
u
p
ar
e
s
h
o
w
n
i
n
Fi
g
u
r
e
9
.
Fig
u
r
e
9
.
Sa
m
p
le
s
to
n
e
tex
t
u
r
e
i
m
ag
e
s
f
r
o
m
v
ar
io
u
s
d
atab
ase
s
,
1
6
f
r
o
m
ea
c
h
clas
s
T
h
e
th
r
ee
C
M
f
ea
t
u
r
es
i.e
C
o
r
r
elatio
n
(
C
R
)
,
C
l
u
s
ter
P
r
o
m
i
n
en
ce
(
C
P
)
an
d
I
n
f
o
r
m
atio
n
m
ea
s
u
r
e
o
f
co
r
r
elatio
n
1
(
I
MC1
)
ar
e
ex
tr
a
cted
o
n
to
t
h
e
DD
R
G
R
M
m
o
d
el
o
f
d
i
f
f
er
en
t
s
to
n
e
tex
tu
r
e
g
r
o
u
p
s
o
f
i
m
a
g
e
s
a
n
d
th
e
r
es
u
lt
s
ar
e
s
to
r
ed
in
t
h
e
f
ea
tu
r
e
v
ec
to
r
.
Feat
u
r
e
s
et
lea
d
s
to
r
ep
r
esen
tatio
n
o
f
tr
ai
n
in
g
i
m
ag
e
s
.
T
h
e
th
r
ee
C
M
f
ea
tu
r
e
s
o
f
s
to
n
e
i
m
ag
e
s
o
f
f
o
u
r
g
r
o
u
p
s
i.e
.
Ma
r
b
le,
Gr
a
n
ite,
B
r
ick
s
a
n
d
Mo
s
aic
ar
e
s
h
o
w
n
i
n
T
ab
les 1
,
2
,
3
,
an
d
4
r
esp
ec
ti
v
el
y
.
B
ased
o
n
t
h
ese
f
ea
tu
r
e
s
et
v
al
u
es
t
h
e
t
ested
i
m
ag
e
is
clas
s
i
f
ied
b
y
u
s
in
g
o
n
e
o
f
th
e
t
w
o
ap
p
r
o
ac
h
es
an
d
class
if
ied
th
e
s
to
n
e
i
m
a
g
es
in
to
o
n
e
o
f
th
e
f
o
u
r
p
r
e
-
d
e
f
i
n
ed
g
r
o
u
p
s
i.e
.
,
Ma
r
b
le,
Gr
an
ite,
B
r
ick
s
an
d
Mo
s
aic.
T
h
e
f
ir
s
t
a
p
p
r
o
ac
h
u
s
es
th
e
s
tan
d
ar
d
clas
s
if
ica
tio
n
al
g
o
r
ith
m
s
a
n
d
s
ec
o
n
d
ap
p
r
o
ac
h
u
s
es
a
u
s
er
d
ef
i
n
ed
alg
o
r
it
h
m
.
T
ab
le
1
.
Featu
r
e
s
et
v
al
u
es o
f
th
e
g
r
a
n
ite
te
x
t
u
r
es
S
n
o
I
M
A
G
E
N
A
M
E
C
M
f
e
a
t
u
r
e
s o
n
D
D
R
G
R
M
mo
d
e
l
C
o
r
r
e
l
a
t
i
o
n
C
l
u
st
e
r
P
r
o
mi
n
e
n
c
e
I
n
f
o
r
mat
i
o
n
me
a
s
u
r
e
o
f
c
o
r
r
e
l
a
t
i
o
n
1
1
G
r
a
n
i
t
e
0
0
1
0
.
0
2
1
3
5
9
3
0
.
7
5
6
3
2
G
r
a
n
i
t
e
0
0
2
0
.
0
3
5
6
6
1
3
2
.
3
6
3
G
r
a
n
i
t
e
0
0
3
0
.
1
0
2
1
6
0
1
1
3
.
9
1
4
G
r
a
n
i
t
e
0
0
4
0
.
0
5
6
4
4
8
7
1
2
.
3
6
5
G
r
a
n
i
t
e
0
0
5
0
.
0
9
6
7
4
7
6
1
0
.
3
5
6
G
r
a
n
i
t
e
0
0
6
0
.
1
0
6
3
5
1
3
9
.
3
6
7
G
r
a
n
i
t
e
0
0
7
0
.
1
0
0
7
5
7
6
8
.
6
4
8
G
r
a
n
i
t
e
0
0
8
0
.
0
9
6
5
5
1
9
7
.
4
9
9
G
r
a
n
i
t
e
0
0
9
0
.
0
6
4
0
4
1
2
6
.
7
6
10
G
r
a
n
i
t
e
0
1
0
0
.
0
4
0
2
4
7
2
2
.
4
2
11
G
r
a
n
i
t
e
0
1
1
0
.
0
5
0
8
3
9
6
4
.
3
6
12
G
r
a
n
i
t
e
0
1
2
0
.
0
1
9
7
3
8
6
0
.
6
6
13
G
r
a
n
i
t
e
0
1
3
0
.
0
2
3
1
3
9
3
0
.
9
9
14
G
r
a
n
i
t
e
0
1
4
0
.
0
9
3
7
5
9
2
1
1
.
7
8
15
G
r
a
n
i
t
e
0
1
5
0
.
0
9
4
1
5
5
3
1
2
.
3
7
16
G
r
a
n
i
t
e
0
1
6
0
.
0
4
5
2
4
1
2
2
.
3
5
17
G
r
a
n
i
t
e
0
1
7
0
.
0
7
3
3
4
0
2
8
.
9
7
18
G
r
a
n
i
t
e
0
1
8
0
.
0
8
2
8
4
9
7
1
0
.
1
6
19
G
r
a
n
i
t
e
0
1
9
0
.
1
0
7
0
7
7
8
1
3
.
0
8
20
G
r
a
n
i
t
e
0
2
0
0
.
0
3
5
4
7
1
2
1
.
3
5
0
0
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
A
N
o
ve
l A
p
p
r
o
a
ch
B
a
s
ed
o
n
Dec
r
ea
s
ed
Dimen
s
io
n
a
n
d
R
e
d
u
ce
d
Gra
y…
(
G
S
N
Mu
r
th
y
)
2509
T
ab
le
2
.
Featu
r
e
s
et
v
al
u
es o
f
th
e
b
r
ick
s
te
x
t
u
r
es
S
n
o
I
M
A
G
E
N
A
M
E
C
M
f
e
a
t
u
r
e
s o
n
D
D
R
G
R
M
mo
d
e
l
C
o
r
r
e
l
a
t
i
o
n
C
l
u
st
e
r
P
r
o
mi
n
e
n
c
e
I
n
f
o
r
mat
i
o
n
me
a
s
u
r
e
o
f
c
o
r
r
e
l
a
t
i
o
n
1
1
B
r
i
c
k
0
0
1
0
.
2
3
5
6
1
1
2
2
0
.
5
9
6
2
B
r
i
c
k
0
0
2
0
.
5
6
8
6
1
6
9
3
0
.
2
9
8
8
3
B
r
i
c
k
0
0
3
0
.
3
5
6
8
1
3
6
8
0
.
1
2
3
5
4
B
r
i
c
k
0
0
4
0
.
3
2
8
6
1
0
7
6
0
.
1
1
3
5
5
B
r
i
c
k
0
0
5
0
.
2
2
8
8
8
0
4
0
.
0
6
1
7
6
B
r
i
c
k
0
0
6
0
.
3
1
5
2
9
1
8
0
.
0
8
6
5
7
B
r
i
c
k
0
0
7
0
.
2
9
6
4
1
0
1
3
0
.
1
0
2
5
8
B
r
i
c
k
0
0
8
0
.
2
5
6
4
1
2
4
6
0
.
1
1
3
5
9
B
r
i
c
k
0
0
9
0
.
2
7
3
7
1
1
6
9
0
.
0
7
5
0
10
B
r
i
c
k
0
0
1
0
0
.
3
1
5
6
1
1
7
5
0
.
1
3
6
5
11
B
r
i
c
k
0
0
1
1
0
.
3
6
4
3
1
0
6
9
0
.
2
1
6
5
12
B
r
i
c
k
0
0
1
2
0
.
3
5
9
4
1
5
2
1
0
.
2
6
4
5
13
B
r
i
c
k
0
0
1
3
0
.
2
4
7
6
1
0
0
9
0
.
2
1
3
5
14
B
r
i
c
k
0
0
1
4
0
.
2
3
6
9
1
0
3
5
0
.
1
9
6
5
15
B
r
i
c
k
0
0
1
5
0
.
2
2
0
0
1
0
8
0
0
.
0
4
9
1
16
B
r
i
c
k
0
0
1
6
0
.
2
6
0
5
1
2
1
2
0
.
0
6
6
2
17
B
r
i
c
k
0
0
1
7
0
.
2
6
5
8
1
4
1
2
0
.
0
6
5
5
18
B
r
i
c
k
0
0
1
8
0
.
2
7
5
4
1
1
2
4
0
.
0
5
6
4
19
B
r
i
c
k
0
0
1
9
0
.
2
3
4
1
1
0
3
6
0
.
0
5
7
4
20
B
r
i
c
k
0
0
2
0
0
.
2
4
5
7
1
2
1
7
0
.
1
8
6
5
T
ab
le
3
.
Featu
r
e
s
et
v
al
u
es o
f
th
e
m
o
s
aic
tex
tu
r
es
S
n
o
I
M
A
G
E
N
A
M
E
C
M
f
e
a
t
u
r
e
s o
n
D
D
R
G
R
M
mo
d
e
l
C
o
r
r
e
l
a
t
i
o
n
C
l
u
st
e
r
P
r
o
mi
n
e
n
c
e
I
n
f
o
r
mat
i
o
n
me
a
s
u
r
e
o
f
c
o
r
r
e
l
a
t
i
o
n
1
1
M
o
sai
c
.
0
0
1
0
.
1
7
6
8
8
9
7
0
.
0
0
9
3
2
M
o
sai
c
.
0
0
2
0
.
1
6
5
8
1
1
3
6
0
.
0
1
6
5
3
M
o
sai
c
.
0
0
3
0
.
2
0
1
3
1
0
3
7
0
.
0
5
0
1
4
M
o
sai
c
.
0
0
4
0
.
1
9
4
7
9
6
7
0
.
0
4
6
5
5
M
o
sai
c
.
0
0
5
0
.
1
8
2
2
8
1
3
0
.
0
3
8
7
6
M
o
sai
c
.
0
0
6
0
.
1
2
3
2
8
1
1
0
.
0
1
7
1
7
M
o
sai
c
.
0
0
7
0
.
2
1
5
9
8
3
0
0
.
0
5
4
2
8
M
o
sai
c
.
0
0
8
0
.
1
8
8
5
9
0
7
0
.
0
3
9
2
9
M
o
sai
c
.
0
0
9
0
.
1
9
3
6
8
1
4
0
.
0
3
6
5
10
M
o
sai
c
.
0
1
0
0
.
1
3
9
5
8
8
9
0
.
0
2
1
3
11
M
o
sai
c
.
0
1
1
0
.
1
2
6
5
9
1
4
0
.
0
2
5
8
12
M
o
sai
c
.
0
1
2
0
.
1
3
6
4
9
2
3
0
.
0
4
1
8
13
M
o
sai
c
.
0
1
3
0
.
1
4
5
7
8
5
4
0
.
0
4
6
8
14
M
o
sai
c
.
0
1
4
0
.
1
5
6
8
8
7
2
0
.
0
4
9
5
15
M
o
sai
c
.
0
1
5
0
.
1
3
5
6
9
6
3
0
.
0
4
6
7
16
M
o
sai
c
.
0
1
6
0
.
1
5
0
8
9
9
8
0
.
0
2
5
4
17
M
o
sai
c
.
0
1
7
0
.
2
1
5
3
9
4
7
0
.
0
5
0
5
18
M
o
sai
c
.
0
1
8
0
.
2
0
5
2
8
6
5
0
.
0
4
7
9
19
M
o
sai
c
.
0
1
9
0
.
0
8
2
2
8
0
9
0
.
0
0
7
3
20
M
o
sai
c
.
0
2
0
0
.
1
6
6
4
9
4
0
0
.
0
2
9
5
T
ab
le
4
.
Featu
r
e
s
et
v
al
u
es o
f
th
e
m
ar
b
le
te
x
t
u
r
es
S
n
o
I
M
A
G
E
N
A
M
E
G
L
C
M
f
e
a
t
u
r
e
s o
n
D
D
R
G
R
M
mo
d
e
l
C
o
r
r
e
l
a
t
i
o
n
C
l
u
st
e
r
P
r
o
mi
n
e
n
c
e
I
n
f
o
r
mat
i
o
n
me
a
s
u
r
e
o
f
c
o
r
r
e
l
a
t
i
o
n
1
1
M
a
r
b
l
e
.
0
0
1
0
.
1
9
5
7
4
4
1
5
7
.
1
8
2
M
a
r
b
l
e
.
0
0
2
0
.
1
3
1
3
7
8
7
1
9
.
9
6
3
M
a
r
b
l
e
.
0
0
3
0
.
1
2
6
7
6
1
3
2
0
.
5
6
4
M
a
r
b
l
e
.
0
0
4
0
.
1
3
6
4
6
4
8
3
0
.
1
2
5
M
a
r
b
l
e
.
0
0
5
0
.
1
4
6
3
5
1
3
3
1
.
1
5
6
M
a
r
b
l
e
.
0
0
6
0
.
1
2
6
6
6
2
4
2
1
.
1
3
7
M
a
r
b
l
e
.
0
0
7
0
.
1
2
3
5
5
2
9
2
1
.
8
5
8
M
a
r
b
l
e
.
0
0
8
0
.
1
3
6
8
5
7
3
2
2
.
8
3
9
M
a
r
b
l
e
.
0
0
9
0
.
1
2
9
9
6
1
3
2
6
.
4
6
10
M
a
r
b
l
e
.
0
10
0
.
1
3
6
2
6
3
8
2
4
.
2
0
11
M
a
r
b
l
e
.
0
1
1
0
.
1
4
7
2
6
7
9
2
4
.
6
5
12
M
a
r
b
l
e
.
0
1
2
0
.
1
3
6
9
7
1
3
2
5
.
6
8
13
M
a
r
b
l
e
.
0
1
3
0
.
1
4
5
8
7
5
2
2
6
.
2
9
14
M
a
r
b
l
e
.
0
1
4
0
.
1
8
6
4
7
4
3
2
7
.
4
9
15
M
a
r
b
l
e
.
0
1
5
0
.
1
7
6
5
4
8
1
2
8
.
1
3
16
M
a
r
b
l
e
.
0
1
6
0
.
1
8
6
1
3
7
8
3
0
.
1
5
17
M
a
r
b
l
e
.
0
1
7
0
.
1
9
5
7
3
8
1
5
9
.
9
8
18
M
a
r
b
l
e
.
0
18
0
.
1
6
1
3
6
0
3
3
4
.
9
9
19
M
a
r
b
l
e
.
0
1
9
0
.
1
8
2
3
4
2
3
5
0
.
8
4
20
M
a
r
b
l
e
.
0
2
0
0
.
1
6
7
1
4
1
3
4
2
.
1
8
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
5
,
Octo
b
er
2
0
1
7
:
2
5
0
2
–
2
5
1
3
2510
3
.
1
.
B
y
us
ing
Sta
nd
a
rd
cla
s
s
if
ica
t
io
n Alg
o
rit
h
m
s
T
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
is
test
ed
u
s
in
g
k
-
Nea
r
est
Nei
g
h
b
o
r
C
las
s
i
f
ier
(
K
-
NN
C
)
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
s
(
SV
M)
ar
e
u
s
ed
f
o
r
class
i
f
icatio
n
p
u
r
p
o
s
e
.
All e
x
p
er
i
m
en
ts
ar
e
ca
r
r
ied
o
u
t o
n
a
P
C
m
ac
h
in
e
w
it
h
i5
p
r
o
ce
s
s
o
r
2
.
6
GHz
C
P
U
an
d
4
GB
R
A
M
m
e
m
o
r
y
u
n
d
er
MA
T
L
A
B
1
0
.
1
a
p
latf
o
r
m
.
40
%
o
f
t
h
e
ea
ch
d
atab
ase
is
u
s
ed
f
o
r
tr
ain
i
n
g
a
n
d
r
e
m
a
i
ni
n
g
6
0
%
i
m
ag
e
s
ar
e
u
s
ed
f
o
r
test
in
g
p
u
r
p
o
s
e
i.e
.
1
0
0
8
im
ag
e
s
ar
e
u
s
ed
f
o
r
tr
ain
i
n
g
p
u
r
p
o
s
e
an
d
1
5
1
4
im
ag
e
s
ar
e
u
s
ed
f
o
r
te
s
ti
n
g
p
u
r
p
o
s
e.
T
h
e
p
er
ce
n
tag
e
o
f
class
i
f
icatio
n
o
f
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
ith
K
-
N
N
C
ap
p
lied
an
d
g
en
er
ated
v
al
u
es
ar
e
lis
ted
o
u
t
i
n
T
ab
le
5
.
T
h
e
p
er
ce
n
tag
e
o
f
class
i
f
icatio
n
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
ith
S
u
p
p
o
r
t
Vec
to
r
Ma
ch
i
n
e
(
SVM
)
ap
p
lied
an
d
g
en
er
ated
v
al
u
es
ar
e
lis
ted
o
u
t i
n
T
ab
le
6
.
T
ab
le
5
.
P
er
ce
n
tag
e
o
f
clas
s
if
i
ca
tio
n
w
h
e
n
k
-
N
NC
al
g
o
r
ith
m
is
ap
p
lied
T
e
x
t
u
r
e
G
r
o
u
p
C
l
a
ssi
f
i
c
a
t
i
o
n
R
a
t
e
o
f
c
o
n
s
i
d
e
r
e
d
S
t
o
n
e
t
e
x
t
u
r
e
D
a
t
a
b
a
se
s w
h
e
n
k
-
NN
c
l
a
ssi
f
i
e
r
a
p
p
l
i
e
d
V
i
sT
e
x
M
a
y
a
n
g
G
o
o
g
l
e
P
a
u
l
B
o
u
r
k
e
O
v
e
r
a
l
l
%
B
r
i
c
k
s
9
5
.
9
9
5
.
7
6
9
5
.
9
8
9
6
.
3
5
96
M
a
r
b
l
e
9
5
.
9
4
9
6
.
3
4
9
6
.
0
6
9
6
.
0
4
9
6
.
1
G
r
a
n
i
t
e
9
5
.
9
9
9
5
.
7
6
9
6
.
2
8
9
6
.
3
5
9
6
.
1
M
o
sai
c
9
5
.
8
8
9
6
.
0
2
9
6
.
5
2
9
5
.
3
9
5
.
9
3
T
ab
le
6
.
P
er
ce
n
tag
e
o
f
clas
s
if
i
ca
tio
n
w
h
e
n
SVM
al
g
o
r
it
h
m
i
s
ap
p
lied
T
e
x
t
u
r
e
G
r
o
u
p
C
l
a
ssi
f
i
c
a
t
i
o
n
R
a
t
e
o
f
c
o
n
s
i
d
e
r
e
d
S
t
o
n
e
t
e
x
t
u
r
e
D
a
t
a
b
a
se
s w
h
e
n
S
V
M
c
l
a
ss
i
f
i
e
r
a
p
p
l
i
e
d
V
i
sT
e
x
M
a
y
a
n
g
G
o
o
g
l
e
P
a
u
l
B
o
u
r
k
e
O
v
e
r
a
l
l
%
B
r
i
c
k
s
9
5
.
9
3
9
6
.
1
3
9
6
.
1
4
9
6
.
0
5
9
6
.
0
6
M
a
r
b
l
e
9
6
.
5
1
9
6
.
2
1
9
6
.
2
4
9
6
.
0
9
9
6
.
2
6
G
r
a
n
i
t
e
9
5
.
9
3
9
6
.
4
3
9
6
.
2
4
9
6
.
1
4
9
6
.
1
9
M
o
sai
c
9
6
.
1
9
9
6
.
6
7
9
6
.
0
7
9
6
.
0
3
9
6
.
2
4
Fro
m
ab
o
v
e
t
w
o
tab
le
s
,
it
i
s
o
b
s
er
v
e
d
th
at
w
h
e
n
t
h
e
K
-
N
N
c
lass
i
f
ier
ap
p
lied
to
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
b
tain
ed
class
i
f
icatio
n
p
er
ce
n
tag
e
a
s
9
6
.
0
3
%
an
d
th
e
cla
s
s
i
f
icatio
n
p
er
ce
n
tag
e
w
h
e
n
SV
M
is
ap
p
lied
i
s
9
6
.
1
9
%
.
A
l
m
o
s
t
t
w
o
cla
s
s
i
f
i
ca
tio
n
al
g
o
r
ith
m
s
ga
v
e
s
a
m
e
class
i
f
icat
io
n
p
er
ce
n
ta
g
e
a
n
d
it
is
h
i
g
h
.
So
t
h
e
p
r
o
p
o
s
ed
DDRGR
M
m
o
d
el
i
s
w
ell
s
u
ited
f
o
r
ex
tr
ac
tio
n
o
f
f
ea
t
u
r
es
f
r
o
m
s
to
n
e
i
m
ag
e
s
an
d
to
class
if
y
th
e
s
to
n
e
te
x
tu
r
e
s
in
to
4
g
r
o
u
p
s
.
3.
2
.
B
y
us
ing
Sta
nd
a
rd
cla
s
s
if
ica
t
io
n Alg
o
rit
h
m
s
B
ased
o
n
t
h
e
f
ea
tu
r
e
s
e
x
tr
ac
te
d
o
n
t
h
e
tr
ai
n
in
g
d
ata
s
et,
th
e
p
r
o
p
o
s
ed
u
s
er
d
ef
in
ed
ap
p
r
o
ac
h
d
er
iv
e
s
a
class
if
ica
tio
n
ap
p
r
o
ac
h
a
s
s
h
o
w
n
i
n
al
g
o
r
ith
m
1
to
clas
s
i
f
y
th
e
s
to
n
e
tex
t
u
r
es
i
n
to
o
n
e
o
f
t
h
e
f
o
u
r
p
r
ed
e
f
i
n
ed
g
r
o
u
p
s
.
So
as
to
te
s
t
t
h
e
ef
f
ic
ien
c
y
o
f
t
h
e
u
s
er
d
ef
i
n
ed
cla
s
s
if
ica
tio
n
ap
p
r
o
ac
h
th
e
tes
t
d
ata
s
et
is
co
llected
r
an
d
o
m
l
y
f
r
o
m
d
i
f
f
er
e
n
t sto
n
e
tex
tu
r
e
d
atab
ases
.
Alg
o
rit
h
m
1
:
A
lg
o
r
it
h
m
f
o
r
C
lass
i
f
icatio
n
o
f
Sto
n
e
tex
t
u
r
es
in
to
4
p
r
e
-
d
ef
in
ed
g
r
o
u
p
s
u
s
i
n
g
C
M
f
ea
t
u
r
e
o
n
DDRG
R
M
m
o
d
el
o
f
s
to
n
e
i
m
ag
es.
B
eg
in
if
C
P
>
8
0
0
&
&
C
R
>=
0
.
2
1
9
th
en
P
r
in
t (
s
to
n
e
i
m
a
g
e
ag
e
i
s
clas
s
if
ied
as
'
B
r
ick
s
C
la
s
s
’
)
;
E
ls
e
if
C
P
>
8
0
0
&
&
C
R
<
0
.
2
1
9
th
en
P
r
in
t (
s
to
n
e
i
m
a
g
e
ag
e
i
s
clas
s
if
ied
as
'
Mo
s
a
ic
C
la
s
s
'
)
;
E
ls
e
if
C
P
<
8
0
0
&
&
I
MC1
>
1
9
th
en
P
r
in
t (
s
to
n
e
i
m
a
g
e
ag
e
i
s
clas
s
if
ied
as
'
Ma
r
b
le
C
la
s
s
’
)
;
E
ls
e
if
C
P
<
8
0
0
&
&
I
MC1
<
1
9
th
en
P
r
in
t (
s
to
n
e
i
m
a
g
e
ag
e
i
s
clas
s
if
ied
as
'
Gr
a
n
ite
C
las
s
'
)
;
E
ls
e
P
r
in
t (
s
to
n
e
i
m
a
g
e
ag
e
i
s
clas
s
if
ied
as
'
U
n
k
o
n
o
w
n
C
la
s
s
'
)
;
E
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
A
N
o
ve
l A
p
p
r
o
a
ch
B
a
s
ed
o
n
Dec
r
ea
s
ed
Dimen
s
io
n
a
n
d
R
e
d
u
ce
d
Gra
y…
(
G
S
N
Mu
r
th
y
)
2511
T
ab
le
7.
C
lass
if
icatio
n
r
ates o
f
s
to
n
e
i
m
ag
e
s
in
to
4
g
r
o
u
p
s
u
s
in
g
C
M
f
ea
tu
r
e
o
n
D
DR
G
R
M
m
o
d
el
o
f
tex
t
u
r
e
i
m
a
g
es b
ased
o
n
A
l
g
o
r
ith
m
1
.
T
e
x
t
u
r
e
G
r
o
u
p
C
l
a
ssi
f
i
c
a
t
i
o
n
R
a
t
e
o
f
c
o
n
s
i
d
e
r
e
d
S
t
o
n
e
t
e
x
t
u
r
e
D
a
t
a
b
a
se
s w
h
e
n
U
se
r
d
e
f
i
n
e
C
l
a
ssi
f
i
c
a
t
i
o
n
A
l
g
o
r
i
t
h
m
i
s
u
se
d
V
i
sT
ex
M
a
y
a
n
g
G
o
o
g
l
e
P
a
u
l
B
o
u
r
k
e
O
v
e
r
a
l
l
%
B
r
i
c
k
s
9
5
.
9
7
96
9
6
.
1
1
9
6
.
2
5
9
6
.
0
8
M
a
r
b
l
e
9
6
.
2
8
9
6
.
3
3
9
6
.
2
9
6
.
1
2
9
6
.
2
3
G
r
a
n
i
t
e
9
6
.
0
1
9
6
.
1
5
9
6
.
3
1
9
6
.
3
9
6
.
1
9
M
o
sai
c
9
6
.
0
9
9
6
.
4
9
6
.
3
5
9
5
.
7
2
9
6
.
1
4
Fro
m
th
e
t
w
o
s
ec
tio
n
s
,
o
b
s
er
v
e
t
h
at
t
h
e
ex
tr
ac
ted
f
ea
t
u
r
es
ar
e
w
ell
s
u
ited
f
o
r
clas
s
i
f
iatio
n
o
f
s
to
n
e
tex
t
u
r
es
w
h
e
n
s
tan
d
ar
d
an
d
u
s
er
d
ef
i
n
ed
class
if
ica
tio
n
al
g
o
r
ith
m
s
.
Fo
r
an
al
y
s
in
g
t
h
e
r
esu
lt
s
th
e
co
n
f
s
io
n
m
atr
i
x
is
g
en
er
ated
w
h
en
u
s
er
d
ef
in
ed
al
g
o
r
ith
m
is
ap
p
lied
o
n
test
d
atab
ase.
T
h
e
co
n
f
u
s
i
o
n
m
atr
i
x
is
s
h
o
wn
in
T
ab
le
8
.
T
h
e
co
n
f
u
s
io
n
m
at
r
ix
s
h
o
w
s
th
e
cla
s
s
i
f
ied
clas
s
f
o
r
ea
ch
in
p
u
t te
x
t
u
r
e
in
test
d
a
tab
ase.
T
ab
le
8
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
T
e
x
t
u
r
e
G
r
o
u
p
M
a
r
b
l
e
M
o
sai
c
B
r
i
c
k
s
G
r
a
n
i
t
e
M
a
r
b
l
e
6
3
2
2
1
1
M
o
sai
c
0
6
2
4
1
1
B
r
i
c
k
s
2
1
6
2
4
2
G
r
a
n
i
t
e
1
2
0
6
2
6
4.
CO
M
P
ARIS
O
N
WI
T
H
O
T
H
E
R
E
X
I
S
T
I
N
G
M
E
T
H
O
D
S:
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
b
ased
o
n
G
L
C
M
h
i
g
h
li
g
h
t
o
n
DDR
GR
M
f
o
r
s
to
n
e
te
x
tu
r
e
cla
s
s
i
f
i
ca
tio
n
h
a
s
s
h
o
w
n
b
etter
class
i
f
icat
io
n
r
at
e
in
co
m
p
ar
is
o
n
w
it
h
o
th
er
e
x
is
t
in
g
ap
p
r
o
ac
h
es.
T
h
e
r
esu
l
ts
o
f
o
th
er
e
x
is
tin
g
ap
p
r
o
ac
h
es
th
at
ar
e
co
n
s
id
er
ed
f
o
r
co
m
p
ar
is
o
n
i
n
cl
u
d
e:
cla
s
s
i
f
icatio
n
ap
p
r
o
ac
h
p
r
o
p
o
s
ed
b
y
Vij
a
y
et
a
l
[
2
2
]
w
h
ic
h
u
s
ed
O
v
er
lap
p
ed
5
-
b
it
T
-
P
atter
n
s
Occ
u
r
r
en
ce
o
n
5
-
by
-
5
s
u
b
i
m
a
g
es
,
W
av
elet
b
ased
His
to
g
r
a
m
o
n
T
ex
to
n
P
atter
n
s
(
W
HT
P)
[
2
3
]
p
r
o
p
o
s
ed
b
y
Sa
s
i
Kir
a
n
et
al
,
tex
t
u
r
e
class
if
icatio
n
b
ased
o
n
T
ex
to
n
Feat
u
r
es
[
2
4
]
b
y
R
a
v
i b
ab
u
et
al
a
n
d
ap
p
r
o
ac
h
b
ased
o
n
S
y
n
tactic
P
at
ter
n
o
n
3
D
tec
h
n
iq
u
e
[
2
5
]
.
I
t is q
u
ite
e
v
id
en
t th
a
t,
th
e
p
r
o
p
o
s
ed
s
tr
ateg
y
r
es
u
lte
d
in
h
i
g
h
c
h
ar
ac
ter
izatio
n
r
at
e
th
an
t
h
e
e
x
is
ti
n
g
tech
n
iq
u
e
s
.
T
h
e
class
if
icatio
n
r
ate
f
o
r
th
e
p
r
o
p
o
s
ed
an
d
o
th
er
ex
is
ti
n
g
s
tr
ate
g
ie
s
ar
e
s
h
o
w
n
i
n
T
ab
le
9
an
d
th
e
s
a
m
e
w
a
s
p
o
r
tr
ay
ed
u
s
i
n
g
g
r
ap
h
ical
r
ep
r
esen
tat
io
n
i
n
F
i
g
u
r
e
1
0
.
T
ab
le
9
.
P
er
ce
n
tag
e
m
ea
n
clas
s
if
ica
tio
n
r
ates
f
o
r
p
r
o
p
o
s
ed
DDRG
R
M
m
o
d
el
an
d
o
th
er
e
x
is
tin
g
m
e
t
h
o
d
s
i
n
th
e
liter
at
u
r
e
I
mag
e
D
a
t
a
b
a
se
5
-
b
i
t
'T
' Pa
t
t
e
r
n
A
p
p
r
o
a
c
h
S
y
n
t
a
c
t
i
c
P
a
t
t
e
r
n
o
n
3
D
me
t
h
o
d
T
e
x
t
o
n
F
e
a
t
u
r
e
D
e
t
e
c
t
i
o
n
W
a
v
e
l
e
t
b
a
se
d
H
i
s
t
o
g
r
a
m
o
n
T
e
x
t
o
n
P
a
t
t
e
r
n
s
P
r
o
p
o
se
d
D
D
R
G
R
M
M
e
t
h
o
d
V
i
sT
e
x
9
5
.
9
5
9
3
.
1
5
9
5
.
4
6
9
2
.
8
7
9
5
.
9
3
T
e
x
t
u
r
e
I
mag
e
s T
a
k
e
n
b
y
C
a
me
r
a
9
6
.
3
5
9
2
.
8
7
9
5
.
1
2
9
1
.
7
9
6
.
8
5
G
o
o
g
l
e
9
6
.
7
6
9
3
.
3
2
9
4
.
8
6
9
3
.
5
6
9
6
.
9
6
M
a
y
a
n
g
9
5
.
8
5
9
2
.
8
3
9
4
.
3
9
9
2
.
9
5
9
6
.
1
5
P
a
u
l
B
o
u
r
k
e
9
5
.
9
3
9
3
.
0
5
9
5
.
2
3
9
3
.
0
5
9
5
.
9
8
Fig
u
r
e
10
.
C
o
m
p
ar
is
o
n
g
r
ap
h
s
h
o
w
i
n
g
t
h
e
cla
s
s
i
f
icatio
n
r
ate
o
f
p
r
o
p
o
s
ed
an
d
o
th
er
ex
is
ti
n
g
ap
p
r
o
ac
h
es
Fo
r
d
if
f
er
en
t d
ata
s
e
ts
89
90
91
92
93
94
95
96
97
VisTe
x
Tex
t
u
re
Im
a
g
e
s
Tak
e
n
b
y
Ca
m
e
ra
Go
o
g
le
M
a
y
a
n
g
P
a
u
l
Bo
u
rk
e
Evaluation Warning : The document was created with Spire.PDF for Python.