T
E
L
KO
M
N
I
KA
T
e
lec
om
m
u
n
icat
ion
,
Com
p
u
t
i
n
g,
E
lec
t
r
on
ics
an
d
Cont
r
ol
Vol.
18
,
No.
4
,
Oc
tober
2020
,
pp
.
2371~2377
I
S
S
N:
1693
-
6930,
a
c
c
r
e
dit
e
d
F
ir
s
t
G
r
a
de
by
Ke
me
nr
is
tekdikti
,
De
c
r
e
e
No:
21/E
/KP
T
/2018
DO
I
:
10.
12928/
T
E
L
KO
M
NI
KA
.
v18i5.
8632
2371
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
A
n
e
f
f
ic
ie
n
t
c
ol
or
i
m
ag
e
c
o
m
p
r
e
ssi
on
t
e
c
h
n
i
q
u
e
Walaa
M
.
Ab
d
-
E
lh
af
iez
1
,
Waj
e
b
Gh
ar
ib
i
2
,
M
oh
am
e
d
Hes
h
m
at
3
1
Facu
l
t
y
o
f
Sc
i
en
ce,
So
h
ag
U
n
i
v
ers
i
t
y
,
E
g
y
p
t
1
Co
l
l
eg
e
o
f
C
o
mp
u
t
er
Sci
e
n
ce
&
In
fo
rma
t
i
o
n
T
ech
n
o
l
o
g
y
,
J
azan
U
n
i
v
ers
i
t
y
,
K
i
n
g
d
o
m
o
f
Sau
d
i
A
ra
b
i
a
2
Sch
o
o
l
o
f
Co
m
p
u
t
i
n
g
a
n
d
E
n
g
i
n
eeri
n
g
,
U
M
K
C
,
MO
.
,
U
SA
3
Facu
l
t
y
o
f
C
o
mp
u
t
er
Sci
e
n
ce
an
d
In
fo
rma
t
i
o
n
S
y
s
t
em,
So
h
ag
U
n
i
v
er
s
i
t
y
,
E
g
y
p
t
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
J
a
n
17,
2018
R
e
vis
e
d
Apr
20,
2020
Ac
c
e
pted
M
a
y
1,
2020
W
e
p
re
s
en
t
a
n
ew
i
mag
e
c
o
mp
re
s
s
i
o
n
me
t
h
o
d
t
o
i
mp
r
o
v
e
v
i
s
u
a
l
p
erce
p
t
i
o
n
o
f
t
h
e
d
eco
m
p
res
s
ed
i
ma
g
es
an
d
ach
i
e
v
e
h
i
g
h
er
i
mag
e
co
mp
res
s
i
o
n
rat
i
o
.
T
h
i
s
met
h
o
d
b
al
a
n
ces
b
e
t
w
ee
n
t
h
e
c
o
mp
re
s
s
i
o
n
rat
e
an
d
i
ma
g
e
q
u
a
l
i
t
y
b
y
co
mp
re
s
s
i
n
g
t
h
e
es
s
en
t
i
a
l
p
ar
t
s
o
f
t
h
e
i
ma
ge
-
ed
g
es
.
T
h
e
k
e
y
s
u
b
j
ec
t
/
e
d
g
e
i
s
o
f
m
o
re
s
i
g
n
i
f
i
can
ce
t
h
a
n
b
ac
k
g
r
o
u
n
d
/
n
o
n
-
e
d
g
e
i
mag
e.
T
a
k
i
n
g
i
n
t
o
co
n
s
i
d
erat
i
o
n
t
h
e
v
a
l
u
e
o
f
i
ma
g
e
c
o
mp
o
n
e
n
t
s
an
d
t
h
e
effect
o
f
s
mo
o
t
h
n
e
s
s
i
n
i
mag
e
c
o
mp
re
s
s
i
o
n
,
t
h
i
s
me
t
h
o
d
c
l
as
s
i
f
i
es
t
h
e
i
ma
g
e
c
o
mp
o
n
e
n
t
s
as
e
d
g
e
o
r
non
-
e
d
g
e.
L
o
w
-
q
u
al
i
t
y
l
o
s
s
y
co
m
p
res
s
i
o
n
i
s
a
p
p
l
i
e
d
t
o
n
o
n
-
ed
g
e
co
m
p
o
n
en
t
s
w
h
erea
s
h
i
g
h
-
q
u
al
i
t
y
l
o
s
s
y
co
m
p
res
s
i
o
n
i
s
a
p
p
l
i
e
d
t
o
ed
g
e
c
o
mp
o
n
e
n
t
s
.
O
u
t
co
me
s
s
h
o
w
t
h
a
t
o
u
r
s
u
g
g
e
s
t
e
d
met
h
o
d
i
s
effi
c
i
en
t
i
n
t
erms
o
f
c
o
mp
re
s
s
i
o
n
rat
i
o
,
b
i
t
s
p
er
-
p
i
x
el
a
n
d
p
ea
k
s
i
g
n
al
t
o
n
o
i
s
e
rat
i
o
.
K
e
y
w
o
r
d
s
:
C
ompr
e
s
s
ion
r
a
ti
o
E
dge
de
tec
ti
on
I
mage
c
ompr
e
s
s
ion
J
P
E
G
L
oc
a
l
thr
e
s
holds
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e
.
C
or
r
e
s
pon
din
g
A
u
th
or
:
W
a
laa
M
.
Abd
-
E
lhaf
iez
,
F
a
c
ult
y
of
S
c
ienc
e
,
S
oha
g
Unive
r
s
it
y,
82524,
S
oha
g,
E
gyp
t.
E
mail:
w_a
_led@
ya
hoo
.
c
om
1.
I
NT
RODU
C
T
I
ON
B
e
c
a
us
e
of
the
a
dva
nc
e
s
in
va
r
ious
a
s
pe
c
ts
of
digi
tal
e
lec
tr
onics
li
ke
im
a
ge
a
c
quis
it
ion,
da
ta
s
tor
a
ge
s
pa
c
e
a
nd
dis
play,
many
ne
w
a
ppli
c
a
ti
ons
of
the
dig
it
a
l
im
a
ging
ha
ve
e
mer
ge
d
withi
n
the
las
t
de
c
a
de
.
H
owe
ve
r
,
s
e
ve
r
a
l
of
thos
e
a
ppli
c
a
ti
ons
don't
s
e
e
m
to
be
wide
s
pr
e
a
d
a
s
a
r
e
s
ult
of
ne
e
de
d
lar
ge
s
pa
c
e
of
s
tor
ing.
C
ons
e
que
ntl
y,
the
im
a
ge
c
ompr
e
s
s
ion
ha
s
gr
own
tr
e
mendous
ly
ove
r
the
las
t
de
c
a
de
a
nd
va
r
iou
s
im
a
ge
co
mpr
e
s
s
ion
a
lgor
it
hms
ha
ve
be
e
n
p
r
opos
e
d
[
1
,
2
]
.
P
ictu
r
e
c
omp
r
e
s
s
ion
r
e
duc
e
s
the
a
mount
of
da
ta
r
e
quir
e
d
to
r
e
p
r
e
s
e
nt
a
digi
tal
im
a
ge
.
T
he
r
e
duc
ti
on
p
r
oc
e
s
s
is
the
r
e
moval
of
unne
c
e
s
s
a
r
y
da
ta.
I
t
ne
e
ds
c
ons
ider
a
ble
a
mount
of
s
tor
a
ge
c
a
pa
c
it
y
a
nd
tr
a
ns
mi
s
s
io
n
ba
nd
width
to
tr
a
ns
f
e
r
mul
t
im
e
dia
mate
r
ial
in
unc
ompr
e
s
s
e
d
f
or
m.
T
his
make
s
tr
a
ns
mi
s
s
ion
s
low
a
nd
ti
me
-
c
on
s
um
ing.
P
hotos
t
r
a
ns
mi
tt
e
d
ove
r
the
W
or
ld
W
ide
W
e
b
a
r
e
a
n
e
xc
e
ll
e
nt
e
xa
mpl
e
of
why
da
ta
c
ompr
e
s
s
ion
is
i
mpor
tant.
C
ompr
e
s
s
ion
c
a
n
be
dis
tr
ibut
e
d
to
los
s
les
s
[
3
,
4
]
or
los
s
y
[
5
,
6]
,
r
e
lyi
ng
on
whe
ther
a
ll
the
in
f
or
mati
on
is
not
gott
e
n
e
li
mi
na
te
o
f
or
s
ome
of
it
is
igno
r
e
d
thr
ough
the
c
ompr
e
s
s
ion
pr
oc
e
s
s
[
7]
.
I
n
the
c
ir
c
ums
tanc
e
of
los
s
les
s
c
ompr
e
s
s
ion,
the
r
e
c
ove
r
e
d
da
ta
is
s
i
mi
lar
to
the
or
igi
na
l
,
a
l
though
,
f
or
los
s
y
c
ompr
e
s
s
ion,
the
r
e
s
tor
e
d
da
ta
is
a
de
tailed
look
-
a
li
ke
o
f
the
or
igi
na
l.
W
he
r
e
ve
r
los
s
les
s
c
ompr
e
s
s
ion
is
int
e
nde
d
f
or
da
ta
li
ke
in
ba
nk
r
e
c
or
ds
,
e
ve
n
a
c
ha
nge
of
a
s
ole
c
ha
r
a
c
ter
c
a
n
b
e
ter
r
ibl
e
.
S
im
il
a
r
ly,
f
or
medic
a
l
or
s
a
telli
te
pictu
r
e
s
,
if
ther
e
is
a
ny
los
s
dur
ing
c
ompr
e
s
s
ion,
it
c
a
n
lea
d
to
a
r
ti
f
a
c
ts
in
the
r
e
c
ons
tr
uc
ti
on
that
may
give
wr
ong
int
e
r
p
r
e
tation.
I
n
los
s
y
c
ompr
e
s
s
ion,
the
a
mount
of
los
s
in
the
da
ta
loca
tes
the
s
tanda
r
d
of
the
r
e
c
ons
tr
uc
ti
on
a
nd
do
e
s
indee
d
not
lea
d
to
c
h
a
nge
in
the
inf
or
mation
c
ontent.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
5
,
Oc
tober
2020:
2371
-
2377
2372
Als
o
it
c
a
n
be
us
e
d
f
or
s
ignals
s
uc
h
a
s
s
pe
e
c
h,
na
tu
r
a
l
im
a
ge
s
.
L
os
s
y
c
ompr
e
s
s
ion
a
c
hieve
d
mor
e
c
ompr
e
s
s
ion
than
los
s
les
s
c
ompr
e
s
s
ion.
J
unc
a
i
Y.
a
nd
Guiz
hong
D.
[
8
]
ha
ve
be
e
n
p
r
e
s
e
nted
a
ne
w
c
olor
im
a
ge
c
ompr
e
s
s
ion
method
us
ing
human
vis
ua
l
dis
ti
nc
ti
on
s
e
ns
it
ivi
ty
c
ha
r
a
c
ter
is
ti
c
s
.
F
ir
s
tl
y,
they
c
onve
r
ted
the
input
im
a
ge
int
o
Y
C
r
C
b
a
nd
divi
de
d
the
im
a
ge
int
o
s
ub
r
e
gions
.
T
he
y
a
ppl
ied
DC
T
f
o
r
e
a
c
h
a
nd
e
ve
r
y
blocks
a
nd
qua
nti
z
a
ti
on.
3
qua
nti
z
a
ti
on
matr
ice
s
ha
ve
buil
t
by
c
omb
ini
ng
the
dis
ti
nc
ti
on
s
e
ns
it
ivi
ty
c
ha
r
a
c
ter
is
ti
c
s
of
human
be
i
ng
vis
ua
l
s
ys
tem.
Af
ter
wa
r
ds
,
they
us
e
d
Huf
f
man
c
ode
.
L
.
S
tar
os
ols
ki,
[
9]
p
r
opos
e
d
e
f
f
e
c
ti
ve
c
olor
s
pa
c
e
c
ha
nge
ment
f
or
los
s
les
s
im
a
ge
c
ompr
e
s
s
ion.
H.
B
.
Ke
kr
e
,
e
t
al
.
[
10]
int
r
oduc
e
d
I
mage
C
omp
r
e
s
s
ion
s
ys
tem
us
in
g
ve
c
tor
qua
nti
z
a
ti
on
a
nd
hybr
id
wa
ve
let
tr
a
ns
f
or
m.
K
r
one
c
ke
r
pr
oduc
t
f
o
r
two
va
r
ious
tr
a
ns
f
or
ms
c
a
n
be
us
e
d
to
c
r
e
a
te
hybr
id
wa
ve
let
t
r
a
ns
f
or
m.
Ali
H
.
Ahme
d
a
nd
L
oa
y
E
.
Ge
or
ge
,
[
11]
pr
e
s
e
nted
c
olor
im
a
ge
c
om
pr
e
s
s
ion
tec
hnique
ba
s
e
d
on
w
a
ve
let,
dif
f
e
r
e
nti
a
l
pu
ls
e
c
ode
modul
a
ti
on
a
nd
qua
dtr
e
e
c
oding.
R
e
c
e
ntl
y,
dif
f
e
r
e
nt
im
a
ge
models
ba
s
e
d
on
f
r
a
c
ti
ona
l
tot
a
l
va
r
iation
ha
ve
b
e
e
n
pr
ovided
[
12]
.
S
pa
c
e
a
nd
wa
ve
let
domain
da
mage
a
r
e
us
e
d
in
the
models
f
or
i
mage
s
with
or
without
nois
e
.
Va
r
ious
f
a
c
tor
s
li
ke
im
a
ge
c
ompr
e
s
s
ion,
im
a
ge
r
e
s
tor
a
ti
on,
im
a
ge
c
oding
a
nd
s
o
on
,
ha
ve
be
e
n
dis
c
us
s
e
d
[
13
-
21]
.
W
it
hin
our
p
r
opos
e
d
method,
f
or
c
olor
im
a
ge
c
o
mpr
e
s
s
ion,
the
e
dge
de
tec
ti
on
a
nd
c
omput
e
r
ize
d
de
r
ivation
of
loca
l
thr
e
s
holds
a
r
e
us
e
d.
T
he
a
lg
or
it
hm
is
c
ompos
e
d
of
3
main
s
tage
s
whe
r
e
the
im
a
ge
is
c
a
tegor
ize
d
us
ing
e
dge
de
tec
ti
on
a
nd
divi
de
d
i
nto
n
×
n
blocks
.
T
he
n
Dis
c
r
e
te
C
os
ine
T
r
a
ns
f
or
m
(
DC
T
)
is
us
e
d
on
the
pa
r
ti
ti
one
d
im
a
ge
with
qua
nti
z
e
d
c
oe
f
f
icie
nts
that
or
de
r
e
d
us
ing
a
da
pti
ve
block
s
c
a
nning.
T
he
v
a
r
ianc
e
/m
e
a
n
a
da
pti
ve
th
r
e
s
hold
will
c
omput
e
to
e
li
mi
na
te
we
a
k
c
oe
f
f
icie
nts
.
I
t
will
r
e
ly
upon
e
a
c
h
c
olor
s
pa
c
e
a
nd
block
s
in
e
a
c
h
c
olor
s
pa
c
e
.
E
xpe
r
im
e
ntal
r
e
s
ult
s
dis
play
a
dva
nc
e
r
e
s
ult
s
in
c
ompr
e
s
s
i
on
r
a
ti
o,
bit
s
pe
r
pixel
a
nd
pe
a
k
s
ignal
to
nois
e
r
a
ti
o
f
or
th
e
r
e
c
ons
tr
uc
ted
im
a
ge
.
T
he
e
f
f
e
c
ti
ve
of
c
ompr
e
s
s
ion
r
a
ti
o
de
pe
nding
on
the
na
tu
r
e
of
the
im
a
ge
f
i
le.
T
he
r
e
s
t
of
our
pa
pe
r
is
or
ga
nize
d
a
s
f
o
ll
ows
.
I
n
s
e
c
ti
on
2
,
w
e
e
xplain
the
c
or
e
pr
oc
e
s
s
that
a
s
s
igns
loca
l
thr
e
s
holds
.
S
e
c
ti
on
3
de
s
c
r
ibes
the
a
da
pti
ve
block
s
c
a
nning
me
thod
a
nd
S
e
c
ti
on
4
pr
e
s
e
nted
the
pr
opos
e
d
im
a
ge
c
ompr
e
s
s
ion
s
tr
a
tegy.
R
e
s
ult
s
a
n
d
dis
c
us
s
ion
a
r
e
given
in
s
e
c
ti
on
5
a
nd
the
pa
pe
r
c
a
me
to
the
c
onc
lus
ion
with
s
e
c
ti
on
6
.
2.
AD
AP
T
I
VE
T
HRE
S
HO
L
D
(
L
OCAL
M
E
A
NS
AN
D
L
OCAL
VA
RI
AN
C
E
S
)
T
h
r
e
s
h
ol
di
ng
te
c
h
ni
qu
e
s
a
r
e
o
f
t
e
n
a
pp
l
ie
d
t
o
s
e
gm
e
n
t
im
a
ge
s
d
iv
id
e
i
no
da
r
k
o
bj
e
c
t
s
a
n
d
br
ig
ht
b
a
c
kg
r
o
un
ds
,
o
r
t
he
ot
he
r
w
a
y
r
ou
nd
.
T
h
is
a
ls
o
o
f
f
e
r
s
da
ta
c
o
mp
r
e
s
s
i
on
a
n
d
f
a
s
t
d
a
ta
p
r
o
c
e
s
s
in
g
[
2
2
,
2
3
]
.
T
h
e
e
a
s
i
e
s
t
wa
y
is
t
h
r
o
ug
h
a
te
c
h
ni
qu
e
c
a
l
le
d
g
lob
a
l
t
h
r
e
s
h
o
ld
in
g
,
w
he
r
e
o
ne
th
r
e
s
ho
ld
v
a
l
ue
is
c
ho
s
e
n
f
o
r
t
he
e
n
ti
r
e
i
ma
ge
w
hi
c
h
is
o
bt
a
i
ne
d
f
r
om
t
he
g
lo
ba
l
in
f
o
r
m
a
t
io
n
.
H
owe
ve
r
,
onc
e
t
he
ba
c
kg
r
ou
nd
h
a
s
n
on
-
u
ni
f
o
r
m
i
l
lu
m
ina
t
io
n
,
a
f
ix
e
d
o
r
gl
ob
a
l
th
r
e
s
ho
ld
va
lue
wi
ll
p
o
or
ly
s
e
g
me
nt
the
im
a
ge
.
T
hus
,
the
va
lue
o
f
lo
c
a
l
t
h
r
e
s
ho
ld
v
a
l
ue
t
ha
t
c
ha
nge
s
d
yn
a
m
ic
a
l
ly
o
ve
r
th
e
im
a
ge
is
r
e
q
u
ir
e
d
.
T
h
is
tec
hn
iq
ue
is
c
a
l
led
a
da
p
t
i
ve
t
h
r
e
s
ho
ld
ing
.
B
e
lo
w
,
w
e
in
t
r
o
du
c
e
a
n
a
ut
o
ma
ti
c
me
th
od
t
ha
t
c
a
lcu
la
tes
a
d
a
p
t
ive
loc
a
l
th
r
e
s
ho
lds
f
o
r
t
he
i
ma
ge
c
om
p
r
e
s
s
io
n
.
T
h
e
e
a
s
y
m
e
th
ods
,
M
e
a
ns
a
n
d
Va
r
i
a
n
c
e
s
a
d
a
p
ti
ve
t
h
r
e
s
h
o
ld
a
r
e
us
e
d
,
w
hi
c
h
i
t
ba
s
e
d
o
n
lo
c
a
l
p
r
o
pe
r
t
ies
o
f
t
he
p
a
r
ts
.
L
e
t
m
(
x
,
y
)
,
th
e
l
oc
a
l
me
a
n
a
t
p
os
it
io
n
(
x
,
y
)
o
f
w
i
nd
ows
s
iz
e
w
×
w
,
m
(
x
,
y
)
c
a
n
b
e
c
om
pu
ted
us
i
ng
t
he
s
u
mm
a
t
io
n
o
ve
r
a
ll
p
ixe
l
v
a
l
ue
s
g
(
i
,
j
)
w
it
hi
n
t
ha
t
w
in
do
w
a
n
d
c
a
n
b
e
w
r
it
te
n
a
s
f
o
l
lows
,
m(
x
,
y
)
=
(
g
(
x
+
w
/
2
,
y
+
w
/2
)
+
g
(
x
-
w
/
2
,
y
-
w
/2
)
–
g
(
x
+
w
/2
,
y
-
w
/
2
)
-
g
(
x
-
w
/2
,
y
+
w
/2
)
)
/
w
2
(
1
)
A
ls
o
,
t
he
c
om
pu
ta
t
io
n
o
f
t
he
lo
c
a
l
va
r
ia
nc
e
v
(
x
,
y
)
[
2
3
]
i
s
de
s
c
r
i
be
d
a
s
:
(
,
)
=
1
√
∑
∑
2
(
,
)
−
2
(
,
)
+
/
2
=
−
/
2
+
/
2
=
−
/
2
(
2
)
3.
AD
AP
T
I
VE
B
L
OCK
S
CA
NN
I
NG
I
n
te
nd
e
d
f
o
r
th
e
a
i
m
to
o
bt
a
i
n
th
e
be
s
t
pos
s
ib
le
c
o
m
pr
e
s
s
io
n
r
a
t
io
(
C
R
)
,
dis
c
r
e
t
e
c
os
ine
t
r
a
ns
f
or
m
(
DC
T
)
h
a
s
b
e
e
n
wi
de
ly
e
m
pl
oye
d
in
im
a
ge
a
n
d
v
ide
o
c
od
i
ng
s
ys
t
e
ms
,
whe
r
e
z
i
gz
a
g
s
c
a
n
is
us
ua
l
l
y
us
e
d
f
o
r
DC
T
c
oe
f
f
i
c
i
e
n
t
o
r
ga
n
iza
ti
on
a
nd
i
t
is
t
he
l
a
s
t
l
e
v
e
l
o
f
p
r
oc
e
s
s
i
ng
a
c
om
pr
e
s
s
e
d
im
a
g
e
in
a
t
r
a
ns
f
o
r
m
c
o
de
r
,
be
f
o
r
e
i
t
is
u
s
e
i
n
f
ina
l
e
n
t
r
o
py
e
nc
od
in
g
s
te
p
.
M
ul
t
ip
le
s
c
a
nn
in
g
s
e
r
v
ic
e
s
a
r
e
be
in
g
us
e
d
(
i
.
e
.
,
ve
r
ti
c
a
l
,
hi
lb
e
r
t
,
z
ig
z
a
g
a
nd
h
o
r
i
z
o
nt
a
l
)
f
or
va
r
io
us
s
p
a
t
ia
l
p
r
e
d
ic
ti
on
di
r
e
c
t
ion
o
n
t
he
bl
oc
k
.
Ho
we
v
e
r
,
d
ue
to
lo
c
a
l
p
r
e
d
ict
i
on
e
r
r
o
r
s
t
he
s
tan
da
r
d
z
ig
z
a
g
s
c
a
n
is
n
o
t
e
f
f
e
c
t
iv
e
a
l
l
ti
me
.
S
o
,
w
e
a
p
pl
y
ou
r
p
r
o
pos
e
d
e
f
f
e
c
t
ive
s
c
a
n
ni
ng
m
e
t
hod
in
[
24
]
w
h
ich
c
e
n
te
r
e
d
on
S
o
r
ti
ng
M
e
t
ho
d
.
I
t
in
c
l
ud
e
s
p
r
o
ve
n
g
oo
d
o
n
i
ma
ge
c
o
mp
r
e
s
s
io
n
r
a
the
r
th
a
n
z
i
gz
a
g
s
c
a
n
.
4.
T
HE
P
ROP
OS
E
D
I
M
AGE
COM
P
RE
S
S
I
ON
A
P
P
ROAC
H
T
h
e
i
np
ut
im
a
ge
is
p
r
i
ma
r
il
y
c
las
s
i
f
ie
d
i
nt
o
e
d
ge
a
nd
n
on
-
e
dg
e
p
or
t
io
ns
us
in
g
C
a
n
ny
e
dg
e
de
te
c
t
or
[
25
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
A
n
e
ff
icie
nt
c
olor
image
c
ompr
e
s
s
ion
tec
hnique
(
W
alaa
M
.
A
bd
-
E
lhaf
iez
)
2373
S
i
nc
e
t
he
C
a
nn
y
e
dge
de
tec
to
r
is
a
s
ig
ni
f
ic
a
n
t
a
n
d
t
r
a
d
it
i
ona
l
ly
us
e
d
c
o
nt
r
ib
ut
i
on
to
e
dg
e
de
te
c
t
io
n
tec
h
n
iq
ue
s
.
A
f
te
r
t
ha
t
t
he
im
a
ge
is
s
u
bd
iv
i
de
d
in
to
8x
8
b
lo
c
ks
a
n
d
DC
T
c
oe
f
f
ic
ien
ts
a
r
e
c
a
l
c
u
la
te
d
f
o
r
e
a
c
h
a
nd
e
ve
r
y
bl
oc
k
.
T
h
e
n
q
ua
nt
iz
a
t
io
n
p
r
oc
e
s
s
is
a
pp
li
e
d
,
w
hic
h
i
t
me
a
ns
r
e
d
uc
in
g
th
e
nu
mb
e
r
of
bi
ts
by
r
e
duc
i
ng
t
he
h
ig
ht
-
f
r
e
qu
e
nc
y
c
oe
f
f
i
c
i
e
n
ts
l
e
a
s
t
i
mp
o
r
t
a
nc
e
t
o
z
e
r
o
.
T
he
qu
a
n
t
iza
t
io
n
is
pe
r
f
o
r
med
c
o
n
f
e
r
r
i
ng
to
qua
n
ti
z
a
t
io
n
tab
le
.
T
h
e
q
ua
nt
ize
d
va
l
ue
s
a
r
e
r
e
a
r
r
a
ng
e
d
r
e
la
ti
ng
t
o
a
da
pt
i
ve
s
c
a
n
s
e
t
up
a
s
de
s
c
r
i
be
d
i
n
s
e
c
t
io
n
3
.
I
f
the
b
l
oc
k
is
c
las
s
i
f
i
e
d
a
s
e
d
ge
o
r
n
on
-
e
dg
e
bl
oc
k
th
e
n
o
ne
c
a
s
e
(
a
o
r
b
)
wi
l
l
be
us
e
d
a
s
de
s
c
r
ib
e
d
i
n
s
te
p
6
.
I
ns
i
de
t
he
f
o
l
lo
wi
ng
t
w
o
met
ho
ds
,
a
va
r
iab
le
t
h
r
e
s
h
o
ld
is
c
r
e
a
te
d
th
a
t
v
a
r
i
e
s
wi
th
b
ot
h
e
a
c
h
c
o
lo
r
s
pa
c
e
(
a
s
de
s
c
r
ibe
i
n
t
he
f
o
l
lo
wi
ng
C
S
me
th
od
)
a
n
d
a
ls
o
in
e
a
c
h
bl
oc
k
in
e
a
c
h
c
o
lo
r
s
pa
c
e
(
a
s
s
e
e
n
in
t
he
ne
xt
DC
S
me
th
od
)
.
A
f
t
e
r
d
is
c
a
r
din
g
m
in
or
c
oe
f
f
i
c
i
e
n
ts
,
t
he
r
e
m
a
i
ni
ng
c
oe
f
f
i
c
i
e
n
ts
a
r
e
c
o
m
pr
e
s
s
e
d
b
y
th
e
Hu
f
f
m
a
n
E
n
c
o
de
r
.
E
nc
od
in
g
c
o
l
or
im
a
ge
is
do
ne
u
s
i
ng
t
h
is
p
r
op
oe
d
me
t
ho
ds
:
−
M
e
th
od
ba
s
e
d
on
c
o
lo
r
s
p
a
c
e
(
C
S
)
:
T
h
e
p
r
o
pos
e
d
c
o
mp
r
e
s
s
io
n
a
lg
or
i
th
m
o
f
C
S
is
c
ons
ti
tu
te
d
o
f
e
ig
ht
ma
in
s
te
ps
t
ha
t
c
o
ul
d
be
s
um
ma
r
ize
d
a
s
f
o
l
low
s
:
S
t
e
p
1:
Ap
pl
y
c
a
nn
y
o
pe
r
a
to
r
f
or
e
d
ge
e
x
tr
a
c
ti
on
on
e
a
c
h
c
o
lo
r
s
pa
c
e
i
ma
ge
.
S
t
e
p
2:
C
om
p
u
te
t
he
a
da
p
t
ive
t
hr
e
s
h
ol
d
(
v
a
r
ia
nc
e
/
me
a
n
)
f
o
r
e
a
c
h
c
o
l
or
s
pa
c
e
t
o
e
li
m
ina
te
we
a
k
c
oe
f
f
i
c
ie
nts
.
S
t
e
p
3:
Di
vi
de
th
e
i
ma
ge
in
to
8x
8
s
u
b
i
ma
ge
s
.
S
t
e
p
4
:
A
pp
ly
D
C
T
on
the
pa
r
t
i
ti
on
e
d
i
ma
ge
(
64
c
oe
f
f
ic
ien
ts
w
il
l
b
e
ob
ta
in
e
d
:
1
DC
c
oe
f
f
ici
e
n
t
a
nd
6
3
A
C
c
o
e
f
f
ic
i
e
nts
.
S
t
e
p
5:
Qua
n
ti
z
e
t
he
c
oe
f
f
ic
ien
ts
.
S
t
e
p
6:
C
las
s
i
f
y
th
e
b
lo
c
ks
to
e
d
ge
a
nd
no
n
-
e
d
ge
blo
c
ks
,
a
n
d
t
he
n
us
e
d
o
ne
c
a
s
e
f
r
o
m
the
f
ol
l
ow
in
g
c
a
s
e
s
:
a.
F
o
r
e
dge
bl
oc
k
,
ma
ke
a
l
l
t
he
c
o
e
f
f
ic
ie
nts
(
les
s
tha
n
a
da
pt
iv
e
va
r
ia
nc
e
th
r
e
s
ho
ld
/
m
o
r
e
tha
n
a
d
a
p
ti
ve
mea
n
th
r
e
s
ho
ld
)
z
e
r
os
.
F
o
r
no
n
-
e
d
ge
bl
oc
k
us
e
d
on
ly
DC
c
oe
f
f
i
c
i
e
n
t
.
b.
F
o
r
e
d
ge
a
n
d
no
n
-
e
d
ge
b
lo
c
k
,
m
a
ke
a
ll
t
he
c
o
e
f
f
ici
e
n
t
s
(
les
s
t
ha
n
a
da
pt
i
ve
va
r
ian
c
e
t
hr
e
s
h
ol
d/
m
or
e
th
a
n
a
da
pt
iv
e
m
e
a
n
th
r
e
s
h
ol
d
)
z
e
r
os
.
S
t
e
p
7:
Or
de
r
th
e
c
o
e
f
f
ic
ie
nts
us
i
ng
z
i
gz
a
g
/ad
a
p
ti
ve
b
l
oc
k
s
c
a
nn
in
g
o
r
de
r
i
ng
(
a
s
i
n
s
e
c
ti
on
3
)
.
S
t
e
p
8:
Ap
pl
y
H
uf
f
ma
n
e
nc
o
di
ng
.
−
M
e
th
od
de
pe
nds
on
bl
oc
ks
i
n
e
a
c
h
c
o
lo
r
s
p
a
c
e
(
DC
S
)
:
T
h
e
a
lg
o
r
i
th
m
o
f
th
e
DC
S
c
a
n
b
e
s
u
m
ma
r
i
z
e
d
a
s
t
he
f
ol
lo
wi
ng
s
te
ps
:
S
t
e
p
1:
Ap
pl
y
c
a
nn
y
o
pe
r
a
to
r
f
or
e
d
ge
e
x
tr
a
c
ti
on
on
e
a
c
h
c
o
lo
r
s
pa
c
e
i
ma
ge
S
t
e
p
2:
Di
vi
de
th
e
i
ma
ge
in
to
8x
8
s
u
b
i
ma
ge
s
.
S
t
e
p
3
:
A
pp
ly
D
C
T
on
the
pa
r
t
i
ti
on
e
d
i
ma
ge
(
64
c
oe
f
f
ic
ien
ts
w
il
l
b
e
ob
ta
in
e
d
:
1
DC
c
oe
f
f
ici
e
n
t
a
nd
6
3
A
C
c
o
e
f
f
ic
ie
nts
.
S
t
e
p
4:
Qua
n
ti
z
e
t
he
c
oe
f
f
ic
ien
ts
.
S
t
e
p
5:
C
o
mp
ut
e
t
he
a
da
p
ti
ve
th
r
e
s
ho
l
d
(
va
r
ia
nc
e
/
mea
n)
f
o
r
e
a
c
h
b
loc
k
in
e
a
c
h
c
ol
o
r
s
p
a
c
e
to
e
l
i
mi
na
te
w
e
a
k
c
o
e
f
f
ic
ie
nts
.
S
t
e
p
6:
C
las
s
i
f
y
th
e
b
lo
c
ks
to
e
d
ge
a
nd
no
n
-
e
d
ge
blo
c
ks
,
a
n
d
t
he
n
us
e
d
o
ne
c
a
s
e
f
r
o
m
the
ne
xt
c
a
s
e
s
:
a.
F
o
r
e
d
ge
b
lo
c
k
,
ma
ke
a
ll
t
he
c
oe
f
f
ic
ie
nt
s
(
m
o
r
e
t
ha
n
(
a
da
p
t
ive
v
a
r
i
a
nc
e
/
me
a
n
t
hr
e
s
h
ol
d
)
)
z
e
r
os
.
A
nd
th
e
n
f
o
r
no
n
-
e
d
ge
bl
oc
k
us
e
d
o
n
ly
D
C
c
o
e
f
f
ic
ie
nt
.
b.
F
o
r
e
dg
e
a
n
d
n
on
-
e
d
ge
b
l
oc
k
,
ma
ke
a
l
l
t
he
c
oe
f
f
ic
ie
nts
(
mo
r
e
t
ha
n
a
d
a
p
ti
ve
(
v
a
r
ia
nc
e
/
mea
n
)
t
h
r
e
s
ho
l
d)
z
e
r
os
.
S
t
e
p
7:
Or
de
r
th
e
c
o
e
f
f
ic
ie
nts
us
i
ng
z
i
gz
a
g
/ad
a
p
ti
ve
b
l
oc
k
s
c
a
nn
in
g
o
r
de
r
i
ng
(
a
s
i
n
s
e
c
ti
on
3
)
.
S
t
e
p
8:
Ap
pl
y
H
uf
f
ma
n
e
nc
o
di
ng
.
T
h
e
d
e
c
o
di
ng
p
r
o
c
e
s
s
is
t
he
in
ve
r
s
o
f
e
nc
od
i
ng
s
c
he
m
e
.
5.
E
XP
E
RI
M
E
NT
AL
RE
S
U
L
T
S
I
n
t
h
is
s
e
c
t
i
on
,
e
x
pe
r
im
e
n
ts
a
r
e
s
ho
wn
t
o
de
mo
ns
t
r
a
te
t
he
p
e
r
f
o
r
ma
nc
e
of
the
p
r
op
os
e
d
im
a
g
e
c
od
in
g
a
pp
r
oa
c
h
.
D
i
f
f
e
r
e
nt
c
o
lo
r
im
a
ge
s
i
n
t
he
R
GB
s
pa
c
e
wi
t
h
d
i
f
f
e
r
e
n
t
c
ha
r
a
c
t
e
r
is
ti
c
s
a
r
e
tes
te
d
i
n
the
e
x
pe
r
im
e
n
ts
i
nc
l
ud
in
g
t
r
e
e
,
b
a
b
oo
n
a
n
d
g
o
ld
h
i
ll
o
f
s
i
z
e
25
6
×
25
6
a
nd
t
r
e
e
2
,
len
a
,
b
a
r
ba
r
a
a
n
d
a
ir
p
lan
e
o
f
s
ize
51
2
×
51
2
.
T
h
e
va
r
i
ous
c
o
m
pr
e
s
s
i
on
m
e
t
ho
ds
s
c
a
n
be
c
o
mp
a
r
e
d
d
e
pe
nd
in
g
on
c
e
r
tai
n
pe
r
f
o
r
ma
nc
e
me
a
s
u
r
e
s
.
C
o
mp
r
e
s
s
i
on
r
a
ti
o
(
C
R
)
is
o
ut
l
ine
d
be
c
a
us
e
t
he
q
ua
nt
it
a
t
iv
e
r
e
l
a
t
io
n
o
f
t
he
qua
nt
i
ty
o
f
b
i
ts
ne
e
de
d
t
o
r
e
p
r
e
s
e
n
t
t
he
i
n
f
o
r
ma
ti
on
b
e
f
o
r
e
c
om
p
r
e
s
s
io
n
t
o
th
e
qu
a
n
ti
t
yr
of
bi
ts
ne
e
d
e
d
o
nc
e
c
om
p
r
e
s
s
io
n
.
R
a
te
is
the
a
ve
r
a
ge
n
um
be
r
o
f
b
it
s
pe
r
s
a
mp
le
o
r
p
ixe
l
(
b
pp
)
,
in
the
ma
tt
e
r
o
f
i
ma
ge
.
D
is
t
o
r
t
i
on
is
q
ua
n
t
if
ie
d
b
y
a
p
a
r
a
m
e
t
e
r
kn
ow
n
a
s
m
e
a
n
s
qua
r
e
e
r
r
o
r
(
M
S
E
)
.
M
S
E
po
i
nts
to
t
he
c
o
mm
on
wo
r
th
o
f
t
he
s
q
ua
r
e
e
r
r
o
r
b
e
t
we
e
n
the
f
i
r
s
t
s
i
gn
a
l
a
n
d
th
e
r
e
f
o
r
e
t
he
r
e
c
o
ns
t
r
uc
ti
on
.
T
h
e
qu
a
l
it
y
o
f
t
he
r
e
c
ons
t
r
u
c
ti
o
n
is
t
he
p
e
a
k
s
ig
na
l
-
to
-
n
ois
e
r
a
t
io
(
P
S
N
R
)
is
in
d
ica
te
d
b
y
t
he
t
op
p
a
r
a
m
e
t
e
r
.
P
S
NR
is
t
he
r
a
t
i
o
of
s
qua
r
e
o
f
the
p
e
a
k
va
lue
o
f
t
he
s
i
gn
a
l
to
t
he
m
e
a
n
s
q
ua
r
e
e
r
r
or
,
s
e
t
b
y
de
c
ib
e
ls
.
5.
1.
Non
-
a
d
ap
t
ive
m
e
t
h
od
A
l
l
f
o
ll
ows
c
a
s
e
s
in
th
is
me
th
od
a
r
e
us
in
g
c
a
nn
y
e
dg
e
d
e
t
e
c
t
i
on
a
nd
z
i
gz
a
g
s
c
a
n
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
5
,
Oc
tober
2020:
2371
-
2377
2374
−
C
a
s
e
1
(
C
S
1)
:
in
th
is
c
a
s
e
,
mea
n
a
da
p
t
ive
t
hr
e
s
h
ol
d
f
o
r
e
a
c
h
c
ol
or
s
pa
c
e
is
c
om
pu
te
d
.
F
o
r
e
d
ge
bl
oc
k
ma
ke
a
l
l
t
he
c
o
e
f
f
ic
ie
nts
(
m
or
e
tha
n
a
da
p
ti
ve
mea
n
th
r
e
s
ho
ld
)
z
e
r
os
.
F
o
r
n
on
-
e
d
ge
bl
oc
k
o
n
ly
DC
c
oe
f
f
ic
ie
nt
i
s
us
e
d
.
−
C
a
s
e
2
(
C
S
2)
:
i
n
th
is
c
a
s
e
,
va
r
i
a
n
c
e
a
d
a
p
ti
ve
t
h
r
e
s
h
o
ld
f
o
r
e
a
c
h
c
o
lo
r
s
p
a
c
e
is
c
om
pu
te
d
.
F
or
e
dg
e
blo
c
k
m
a
k
e
a
l
l
t
h
e
c
o
e
f
f
ic
ie
nts
(
les
s
t
ha
n
a
da
p
t
ive
va
r
ia
nc
e
t
h
r
e
s
h
o
ld
)
z
e
r
os
.
F
o
r
n
on
-
e
dg
e
b
l
oc
k
o
nl
y
D
C
c
oe
f
f
i
c
i
e
n
t
is
us
e
d
.
−
C
a
s
e
3
(
C
S
3
)
:
in
t
his
c
a
s
e
,
v
a
r
ia
nc
e
a
da
p
t
ive
t
h
r
e
s
h
o
ld
f
o
r
e
a
c
h
c
ol
or
s
p
a
c
e
is
c
o
m
pu
ted
.
F
o
r
e
dg
e
a
n
d
non
-
e
d
ge
b
l
oc
ks
,
ma
ke
a
l
l
t
he
c
oe
f
f
ici
e
n
ts
(
l
e
s
s
t
ha
n
a
da
pt
iv
e
va
r
ia
nc
e
th
r
e
s
h
ol
d
)
z
e
r
os
.
−
C
a
s
e
4
(
DC
S
1
)
:
in
t
his
c
a
s
e
,
m
e
a
n
a
d
a
p
ti
ve
t
h
r
e
s
h
o
ld
f
o
r
e
a
c
h
b
loc
k
in
e
a
c
h
c
o
lo
r
s
pa
c
e
is
c
om
put
e
d
.
F
o
r
e
d
ge
bl
oc
k
m
a
k
e
a
ll
t
he
c
oe
f
f
ic
ien
ts
(
mo
r
e
t
ha
n
a
da
p
t
ive
m
e
a
n
t
h
r
e
s
h
o
ld
)
z
e
r
os
.
F
or
n
on
-
e
dg
e
b
lo
c
k
on
ly
DC
c
o
e
f
f
ic
ie
nt
is
us
e
d
.
−
C
a
s
e
5
(
DC
S
2
)
:
in
t
his
c
a
s
e
,
m
e
a
n
a
d
a
p
ti
ve
t
h
r
e
s
h
o
ld
f
o
r
e
a
c
h
b
loc
k
in
e
a
c
h
c
o
lo
r
s
pa
c
e
is
c
om
p
ut
e
d
.
F
o
r
e
dg
e
a
n
d
n
on
-
e
d
ge
b
l
oc
ks
,
ma
ke
a
l
l
t
he
c
oe
f
f
i
c
ie
nts
(
mo
r
e
t
ha
n
a
d
a
p
ti
ve
mea
n
t
h
r
e
s
ho
ld
)
z
e
r
os
.
−
C
a
s
e
6
(
DC
S
3
)
:
i
n
th
is
c
a
s
e
,
va
r
ia
nc
e
a
da
pt
iv
e
th
r
e
s
ho
ld
f
o
r
e
a
c
h
b
l
oc
k
in
e
a
c
h
c
ol
o
r
s
pa
c
e
is
c
o
mp
ute
d
.
F
o
r
e
d
ge
bl
oc
k
m
a
ke
a
l
l
th
e
c
oe
f
f
i
c
i
e
n
ts
(
m
o
r
e
th
a
n
a
da
pt
iv
e
v
a
r
ia
nc
e
t
h
r
e
s
ho
ld
)
z
e
r
os
.
F
or
n
on
-
e
dg
e
b
lo
c
k
o
n
ly
DC
c
o
e
f
f
ic
ie
nt
is
us
e
d
.
−
C
a
s
e
7
(
DC
S
4
)
:
i
n
th
is
c
a
s
e
,
va
r
ia
nc
e
a
da
pt
iv
e
th
r
e
s
ho
ld
f
o
r
e
a
c
h
b
l
oc
k
in
e
a
c
h
c
ol
o
r
s
pa
c
e
is
c
o
mp
ute
d
.
F
o
r
e
dg
e
a
n
d
n
on
-
e
d
ge
b
l
oc
ks
,
ma
ke
a
l
l
t
he
c
oe
f
f
i
c
ie
nts
(
mo
r
e
t
ha
n
a
d
a
p
ti
ve
va
r
i
a
n
c
e
t
h
r
e
s
ho
ld
)
z
e
r
os
.
T
h
e
a
na
l
ys
is
f
a
c
to
r
s
o
f
p
r
o
pos
e
d
n
on
-
a
da
pt
i
ve
me
th
od
on
di
f
f
e
r
e
n
t
i
ma
ge
s
a
r
e
gi
ve
n
i
n
T
a
b
le
1
.
T
h
e
r
e
s
u
lt
s
s
how
,
th
e
ut
il
iz
a
t
io
n
o
f
va
r
ia
nc
e
t
h
r
e
s
ho
l
d
f
or
e
a
c
h
bl
oc
k
in
e
a
c
h
c
ol
o
r
s
pa
c
e
(
DC
S
4
a
n
d
D
C
S
4
-
4
)
h
a
s
i
nc
r
e
a
s
e
th
e
C
R
w
hi
le
p
r
e
s
e
r
vi
ng
th
e
i
ma
ge
q
ua
l
i
ty
.
T
a
ble
1
.
C
ompr
e
s
s
ion
r
a
ti
o,
bit
r
a
te
a
nd
ps
nr
va
lues
a
tt
a
ined
by
non
-
a
da
pti
ve
method
I
M
A
G
E
C
S
1
C
S
2
C
S
3
D
C
S
1
D
C
S
2
D
C
S
3
D
C
S
4
L
E
N
A
P
S
N
R
35.572
35.86
36.32
35.565
35.92
33.35
33.375
CR
19.160
16.63
14.97
18.656
17.44
41.55
41.347
BP
P
1.252
1.442
1.602
1.286
1.375
0.577
0.580
F
RU
IT
P
S
N
R
34.952
35.21
35.61
34.935
35.23
33.16
33.186
CR
19.170
15.93
14.33
17.970
16.60
36.09
35.401
BP
P
1.251
1.506
1.674
1.335
1.445
0.664
0.677
B
A
B
O
O
N
P
S
N
R
31.275
31.59
31.78
31.239
31.38
29.70
29.714
CR
9.052
7.047
6.765
8.430
8.169
35.88
35.716
BP
P
2.651
3.405
3.547
2.847
2.937
0.668
0.6720
A
IRP
L
A
N
E
P
S
N
R
35.573
36.84
37.08
35.969
36.06
33.24
33.244
CR
22.248
15.45
14.20
19.053
18.65
44.52
44.410
BP
P
1.078
1.552
1.689
1.259
1.286
0.539
0.540
5.
2.
Adap
t
ive
m
e
t
h
od
E
v
e
r
y
f
o
l
lo
ws
c
a
s
e
s
i
n
t
hi
s
me
t
ho
d
a
r
e
u
s
i
ng
c
a
n
ny
e
dg
e
de
te
c
t
io
n
a
nd
a
da
pt
i
ve
s
c
a
n
:
−
C
a
s
e
1
(
AC
S
1
)
:
i
n
t
his
c
a
s
e
,
me
a
n
a
da
pt
iv
e
t
h
r
e
s
ho
ld
f
o
r
e
a
c
h
c
o
l
or
s
pa
c
e
is
c
o
m
pu
ted
.
F
o
r
e
d
ge
bl
oc
k
m
a
k
e
a
l
l
t
he
c
o
e
f
f
ic
ie
nts
(
m
or
e
tha
n
a
da
p
ti
ve
mea
n
th
r
e
s
ho
ld
)
z
e
r
os
.
F
o
r
n
on
-
e
d
ge
bl
oc
k
o
n
ly
DC
c
oe
f
f
ic
ie
nt
i
s
us
e
d
.
−
C
a
s
e
2
(
AC
S
2
)
:
i
n
t
his
c
a
s
e
,
l
oc
a
l
v
a
r
i
a
nc
e
f
o
r
e
a
c
h
c
ol
o
r
s
p
a
c
e
is
c
o
mp
ute
d
,
a
nd
f
o
r
e
d
ge
bl
oc
k
m
a
k
e
a
l
l
t
he
c
oe
f
f
i
c
i
e
n
ts
(
les
s
tha
n
a
d
a
p
ti
ve
v
a
r
ia
nc
e
th
r
e
s
h
o
ld
)
z
e
r
os
.
F
o
r
n
on
-
e
d
ge
b
loc
k
on
l
y
DC
c
o
e
f
f
ic
ie
nt
i
s
us
e
d
.
−
C
a
s
e
3
(
A
C
S
3
)
:
i
n
th
is
c
a
s
e
,
loc
a
l
v
a
r
ia
nc
e
f
o
r
e
a
c
h
c
ol
o
r
s
pa
c
e
is
c
o
mp
ut
e
d
.
F
o
r
e
dg
e
a
n
d
no
n
-
e
d
ge
b
loc
ks
m
a
k
e
a
ll
th
e
c
o
e
f
f
ic
ie
nts
(
les
s
th
a
n
a
da
pt
i
ve
v
a
r
ia
nc
e
th
r
e
s
h
ol
d
)
z
e
r
os
.
−
C
a
s
e
4
(
A
DC
S
1
)
:
i
n
th
is
c
a
s
e
,
m
e
a
n
a
da
p
t
ive
t
h
r
e
s
ho
ld
f
o
r
e
a
c
h
b
lo
c
k
i
n
e
a
c
h
c
ol
o
r
s
p
a
c
e
is
c
om
put
e
d
.
F
o
r
e
d
ge
bl
oc
k
m
a
k
e
a
ll
t
he
c
oe
f
f
ic
ien
ts
(
mo
r
e
t
ha
n
a
da
p
t
ive
m
e
a
n
t
h
r
e
s
h
o
ld
)
z
e
r
os
.
F
or
n
on
-
e
dg
e
b
lo
c
k
o
n
ly
DC
c
o
e
f
f
ic
ie
nt
is
us
e
d
.
−
C
a
s
e
5
(
A
DC
S
2
)
:
i
n
th
is
c
a
s
e
,
m
e
a
n
a
da
p
t
ive
t
h
r
e
s
ho
ld
f
o
r
e
a
c
h
b
lo
c
k
i
n
e
a
c
h
c
ol
o
r
s
p
a
c
e
is
c
om
put
e
d
.
F
o
r
e
dg
e
a
n
d
n
on
-
e
d
ge
b
l
oc
ks
ma
ke
a
l
l
t
he
c
oe
f
f
i
c
i
e
n
ts
(
mo
r
e
th
a
n
a
da
pt
i
ve
m
e
a
n
t
h
r
e
s
h
ol
d
)
z
e
r
os
.
−
C
a
s
e
6
(
AD
C
S
3
)
:
in
th
is
c
a
s
e
,
loc
a
l
va
r
ia
nc
e
f
or
e
a
c
h
b
loc
k
in
e
a
c
h
c
ol
o
r
s
pa
c
e
is
c
om
pu
te
d
.
F
o
r
e
dg
e
b
lo
c
k
m
a
k
e
a
ll
t
he
c
oe
f
f
ic
ie
nts
(
mo
r
e
t
ha
n
a
da
pt
iv
e
v
a
r
i
a
n
c
e
th
r
e
s
h
ol
d
)
z
e
r
os
.
F
o
r
no
n
-
e
d
ge
b
loc
k
on
ly
D
C
c
oe
f
f
i
c
i
e
n
t
is
us
e
d
.
−
C
a
s
e
7
(
AD
C
S
4
)
:
in
th
is
c
a
s
e
,
lo
c
a
l
va
r
ia
nc
e
f
or
e
a
c
h
b
lo
c
k
in
e
a
c
h
c
ol
o
r
s
pa
c
e
is
c
o
m
pu
te
d
.
F
o
r
e
dg
e
a
n
d
non
-
e
d
ge
bl
oc
k
s
m
a
ke
a
l
l
the
c
o
e
f
f
ic
ie
nts
(
mo
r
e
th
a
n
a
d
a
p
ti
ve
va
r
i
a
n
c
e
th
r
e
s
ho
ld
)
z
e
r
os
.
T
a
b
le
2
s
ho
ws
t
he
p
r
o
pos
e
d
me
th
od
pe
r
f
o
r
man
c
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
A
n
e
ff
icie
nt
c
olor
image
c
ompr
e
s
s
ion
tec
hnique
(
W
alaa
M
.
A
bd
-
E
lhaf
iez
)
2375
T
h
e
r
e
c
ons
t
r
u
c
t
e
d
im
a
ge
s
a
r
e
s
ho
wn
in
F
ig
ur
e
1
.
T
he
f
ou
r
c
u
r
v
e
s
a
s
s
ho
wn
in
F
ig
ur
e
2
a
n
d
F
i
gu
r
e
3
a
r
e
d
e
m
ons
t
r
a
te
th
a
t
AC
S
3
c
o
mp
r
e
s
s
i
on
pe
r
f
or
ma
nc
e
is
h
ig
he
r
t
ha
n
C
S
3
c
o
mp
r
e
s
s
i
on
pe
r
f
or
ma
nc
e
.
V
a
r
i
ous
c
om
pa
r
is
o
ns
h
a
v
e
r
e
c
e
n
tl
y
b
e
e
n
pe
r
f
or
me
d
to
pr
ove
t
he
e
f
f
e
c
t
iv
e
ne
s
s
o
f
th
e
p
r
e
s
e
n
te
d
m
e
t
ho
do
lo
gy
ov
e
r
o
th
e
r
s
im
il
a
r
me
th
ods
[
2
6
]
f
or
c
o
lo
r
i
ma
ge
c
o
mp
r
e
s
s
i
on
,
a
s
i
n
T
a
b
le
3
.
T
h
e
r
e
s
ul
ts
s
h
ow
t
ha
t
c
o
mp
r
e
s
s
i
on
r
a
t
i
o
of
i
m
a
ge
s
a
r
e
i
mp
r
o
ve
d
.
T
h
e
qua
nt
i
ty
o
f
im
p
r
o
ve
men
t
i
s
de
p
e
n
de
nt
g
r
e
a
t
ly
on
t
he
na
tu
r
e
o
f
th
e
im
a
ge
;
f
or
i
m
a
ge
s
w
i
th
li
t
tl
e
n
on
-
e
dg
e
b
lo
c
ks
,
s
uc
h
a
s
B
a
bo
on
i
ma
ge
,
th
e
i
m
pr
ov
e
m
e
n
t
i
s
le
s
s
s
i
gn
i
f
i
c
a
n
t
,
ho
we
v
e
r
f
o
r
i
mag
e
s
w
i
th
a
lo
t
of
no
n
-
e
dge
bl
oc
ks
,
t
he
im
p
r
o
ve
me
nt
a
r
e
s
i
gn
i
f
i
c
a
nt
.
T
a
ble
2
.
C
ompr
e
s
s
ion
r
a
ti
o
,
bit
r
a
te
a
nd
ps
nr
va
lues
a
tt
a
ined
by
a
da
pti
ve
method
I
ma
ge
A
C
S
1
A
C
S
2
A
C
S
3
A
D
C
S
1
A
D
C
S
2
A
D
C
S
3
A
D
C
S
4
L
e
na
P
S
N
R
35.572
35.864
36.320
35.5639
35.9235
33.359
33.376
CR
22.198
21.286
19.653
21.1913
19.9568
37.768
37.590
bpp
1.0812
1.1275
1.2211
1.1325
1.2026
0.6355
0.638
H
ous
e
c
P
S
N
R
33.416
34.656
35.047
33.8916
34.1028
31.629
31.636
CR
17.068
15.404
14.578
15.2232
14.7152
30.115
29.994
bpp
1.406
1.5580
1.6463
1.576
5
1.6310
0.7969
0.800
T
r
e
e
P
S
N
R
32.368
32.828
33.071
32.4770
32.6778
30.883
30.892
CR
13.972
13.102
12.221
12.7070
11.8170
24.972
24.773
bpp
1.717
1.8318
1.963
1.8887
2.0310
0.9610
0.968
B
a
boon
P
S
N
R
31.274
31.591
31.780
31.238
31.379
29.705
29.716
CR
12.026
11.115
10.724
10.508
10.165
30.395
30.274
bpp
1.995
2.159
2.23
2.283
2.361
0.789
0.792
A
ir
pl
a
ne
P
S
N
R
35.571
36.854
37.101
35.967
36.065
33.241
33.244
CR
21.762
19.658
18.545
19.983
19.578
39.167
39.085
B
pp
1.102
1.220
1.294
1.201
1.225
0.6
12
0.614
F
igur
e
1
.
T
he
c
ompr
e
s
s
e
d
im
a
ge
s
us
ing
the
pr
opos
e
d
a
da
pti
ve
method
(
a)
(
b)
F
i
gu
r
e
2
.
G
r
a
p
hic
a
l
a
na
l
ys
is
o
f
b
it
r
a
te
(
bp
p
)
vs
ps
n
r
w
i
th
(
a
)
le
na
a
nd
(
b)
t
r
e
e
i
mag
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
5
,
Oc
tober
2020:
2371
-
2377
2376
(
a
)
(
b)
F
i
gu
r
e
3
.
G
r
a
p
hic
a
l
a
na
l
ys
is
o
f
ps
n
r
vs
c
om
p
r
e
s
s
io
n
r
a
ti
o
w
it
h
(
a
)
len
a
a
nd
(
b
)
t
r
e
e
i
mag
e
T
a
ble
3.
C
ompar
is
on
P
S
NR
(
DB
)
,
C
R
a
nd
B
P
P
of
the
pr
opos
e
d
methods
(
DC
S
3
a
nd
AD
C
S
)
a
nd
other
method
R
e
f
. [
26]
P
r
opos
e
d me
th
od (
A
D
C
S
3)
P
r
opos
e
d me
th
od (
D
C
S
3)
I
ma
ge
P
S
N
R
CR
bpp
P
S
N
R
CR
bpp
P
S
N
R
CR
bpp
A
ir
pl
a
ne
34.39
43
37.787
0.6351
33.2417
39.1679
0.6127
33.2421
44.5202
0.539
B
a
boon
30.4870
20.812
1.1532
29.7059
30.3953
0.7896
29.704
35.883
0.668
le
na
33.9259
36.437
0.6587
33.3598
37.7684
0.6355
33.3587
41.5566
0.577
T
r
e
e
2
32.0397
27.820
0.8627
31.6790
28.0483
0.8
557
-
-
-
H
ous
e
33.5263
36.485
0.6578
-
-
-
33.3785
44.8697
0.534
6.
CONC
L
USI
ON
T
h
r
ou
gh
th
is
w
o
r
k
,
a
ne
w
w
a
y
of
c
o
lo
r
im
a
ge
c
om
p
r
e
s
s
io
n
is
p
r
o
pos
e
d
us
i
ng
a
da
p
t
iv
e
c
o
mp
ut
e
r
iz
e
d
d
e
r
i
va
ti
on
of
lo
c
a
l
th
r
e
s
h
ol
ds
.
T
he
s
ug
ge
s
t
e
d
a
pp
r
oa
c
h
i
s
ba
s
e
d
u
po
n
a
d
a
p
ti
ve
t
h
r
e
s
ho
ld
c
om
pu
ta
t
io
n
to
r
e
m
ove
w
e
a
k
c
oe
f
f
i
c
ie
n
ts
.
O
u
r
a
pp
r
oa
c
h
is
de
c
om
pos
e
d
i
nt
o
s
e
ve
r
a
l
c
a
s
e
s
w
it
h
di
f
f
e
r
e
n
t
p
a
r
a
met
e
r
s
.
T
h
e
s
e
t
ype
s
o
f
c
a
s
e
s
b
a
s
e
d
on
a
p
pl
yi
ng
lo
w
qu
a
l
it
y
l
oos
y
c
o
m
pr
e
s
s
i
on
on
n
on
-
e
dg
e
a
r
e
a
s
a
nd
hi
gh
qua
l
it
y
l
oos
e
y
c
om
p
r
e
s
s
io
n
o
n
e
d
ge
pa
r
ts
o
f
im
a
g
e
s
.
O
ut
c
o
mes
s
ho
w
the
i
mp
r
ov
e
m
e
n
t
o
f
a
da
pt
iv
e
m
e
t
ho
d
ove
r
t
he
no
n
-
a
d
a
p
ti
ve
me
t
hod
i
n
q
ua
nt
i
ta
ti
ve
P
S
NR
te
r
ms
a
nd
v
e
r
y
p
a
r
t
ic
ul
a
r
ly
in
v
is
ua
l
qu
a
l
i
ty
o
f
the
r
e
c
o
ns
t
r
uc
te
d
i
mag
e
s
.
As
a
f
u
tur
e
w
o
r
k
,
w
e
w
i
ll
i
mp
le
men
t
th
e
p
r
op
os
e
d
a
pp
r
o
a
c
h
o
n
c
lut
t
e
r
ba
c
k
g
r
o
un
d
i
ma
ge
s
o
r
in
c
a
s
e
wh
e
r
e
the
s
u
bj
e
c
t
ha
s
a
mo
no
-
te
xt
u
r
e
a
nd
mo
no
-
c
o
lo
r
w
hi
le
t
he
b
a
c
k
g
r
o
un
d
ha
s
c
om
pl
ic
a
t
e
d
tex
t
ur
e
s
a
nd
c
ol
o
r
s
.
RE
F
E
RE
NC
E
S
[
1]
Z
h
e
-
Mi
n
g
L
u
,
H
u
i
Pei
,
"
H
y
b
r
i
d
Ima
g
e
Co
m
p
res
s
i
o
n
Sc
h
eme
Bas
e
d
o
n
PV
Q
an
d
D
C
T
V
Q
,
"
IE
ICE
-
Tr
a
n
s
a
ct
i
o
n
s
on
In
f
o
r
m
a
t
i
o
n
a
n
d
S
ys
t
em
s
A
r
c
h
i
ve
,
v
o
l
.
E
8
8
-
D
,
n
o.
10
,
p
p
.
2
4
2
2
-
2
4
2
6
2
0
0
6
.
[
2]
Mart
a
Mrak
,
So
n
j
a
G
rg
i
c,
an
d
M
i
s
l
av
G
rg
i
c,
"
Pi
ct
u
re
Q
u
a
l
i
t
y
Meas
u
res
i
n
Ima
g
e
Co
m
p
res
s
i
o
n
S
y
s
t
ems
,
"
IE
E
E
E
U
R
O
C
O
N
,
2
0
0
3
.
[3
]
D
av
i
d
Sal
o
mo
n
,
"
D
at
a
Co
m
p
res
s
i
o
n
,
Co
m
p
l
e
t
e
Referen
ce,
"
Spri
n
g
e
r
V
er
l
a
g
New
Yo
r
k
,
2
0
0
7
.
[4
]
X
i
w
en
O
w
en
Z
h
ao
,
Z
h
i
h
ai
H
e
n
r
y
H
e,
"
L
o
s
s
l
es
s
Imag
e
Co
mp
res
s
i
o
n
U
s
i
n
g
Su
p
er
-
Sp
a
t
i
al
St
r
u
ct
u
re
Pred
i
c
t
i
o
n
,
"
IE
E
E
S
i
g
n
a
l
P
r
o
ces
s
i
n
g
Let
t
er
s
,
v
o
l
.
1
7
,
n
o
.
4
,
p
p
.
3
8
3
-
386
,
2
0
1
0
.
[5
]
E
d
d
i
e
Bat
i
s
t
a
d
e
L
i
ma
Fi
l
h
o
,
E
d
u
ar
d
o
A
.
B.
d
a
Si
l
v
a
Mu
ri
l
o
Bres
ci
a
n
i
d
e
Carv
al
h
o
,
an
d
Fred
eri
co
Si
l
v
a
Pi
n
ag
é,
"
U
n
i
v
ers
a
l
Imag
e
Co
mp
res
s
i
o
n
U
s
i
n
g
Mu
l
t
i
s
cal
e
Rec
u
rren
t
Pat
t
ern
s
W
i
t
h
A
d
a
p
t
i
v
e
Pro
b
ab
i
l
i
t
y
Mo
d
e
l
,
"
IE
E
E
Tr
a
n
s
a
c
t
i
o
n
s
o
n
Im
a
g
e
P
r
o
ce
s
s
i
n
g
,
v
o
l
.
17,
n
o
.
4
,
p
p
.
5
1
2
-
5
2
7
,
2
0
0
8
.
[6
]
X
i
n
L
i
,
Mi
c
h
ael
T
.
O
rch
ar
d
,
"
Edge
-
D
i
rec
t
ed
Pre
d
i
c
t
i
o
n
fo
r
L
o
s
s
l
es
s
Co
mp
re
s
s
i
o
n
o
f
N
a
t
u
ra
l
Imag
e
s
,
"
IE
E
E
Tr
a
n
s
a
c
t
i
o
n
s
o
n
Im
a
g
e
P
r
o
ce
s
s
i
n
g
,
v
o
l
.
1
0
,
n
o
.
6
,
p
p
.
8
1
3
-
8
1
7
,
2
0
0
1
.
[7
]
K
.
Say
o
o
d
,
"
In
t
r
o
d
u
ct
i
o
n
t
o
D
at
a
Co
m
p
res
s
i
o
n
,
"
H
a
r
c
o
u
r
t
In
d
i
a
P
r
i
v
a
t
e
Li
m
i
t
ed
,
N
e
w
D
e
l
h
i
,
2
n
d
ed
i
t
i
o
n
.
2
0
0
0
.
[8
]
Ju
n
cai
Y
a
o
an
d
G
u
i
zh
o
n
g
L
i
u
,
"
A
n
o
v
e
l
co
l
o
r
i
ma
g
e
co
mp
res
s
i
o
n
a
l
g
o
ri
t
h
m
u
s
i
n
g
t
h
e
h
u
man
v
i
s
u
a
l
co
n
t
r
as
t
s
en
s
i
t
i
v
i
t
y
ch
arac
t
eri
s
t
i
cs
,
"
Ph
o
t
o
n
i
c
Sen
s
o
r
,
v
o
l
.
7
,
n
o
.
1
,
p
p
.
7
2
-
8
1
,
2
0
1
7
.
[9
]
R.
St
aro
s
o
l
s
k
i
,
"
N
e
w
s
i
mp
l
e
a
n
d
eff
i
ci
e
n
t
co
l
o
r
s
p
ace
t
r
an
s
f
o
rmat
i
o
n
s
f
o
r
l
o
s
s
l
es
s
i
ma
g
e
c
o
mp
re
s
s
i
o
n
,
"
Jo
u
r
n
a
l
o
f
V
i
s
u
a
l
Co
m
m
u
n
i
ca
t
i
o
n
a
n
d
Im
a
g
e
R
e
p
r
e
s
en
t
a
t
i
o
n
,
v
o
l
.
2
5
,
n
o
.
5
,
p
p
.
1
0
5
6
-
1
0
6
3
,
2
0
1
4
.
[1
0
]
H.
B.
K
ek
re,
Prac
h
i
N
at
u
,
T
a
n
u
j
a
Saro
d
e,
"
Co
l
o
r
Imag
e
Co
mp
re
s
s
i
o
n
U
s
i
n
g
V
ect
o
r
Q
u
an
t
i
za
t
i
o
n
a
n
d
H
y
b
r
i
d
W
a
v
el
e
t
T
ran
s
fo
rm,
"
P
r
o
ced
i
a
C
o
m
p
u
t
er
S
c
i
en
ce
,
v
o
l
.
8
9
,
p
p
.
778
-
7
8
4
,
2
0
1
6
.
[1
1
]
A
l
i
H
.
,
A
h
me
d
an
d
L
o
ay
E
.
G
eo
rg
e,
"
Res
earch
A
rt
i
cl
e
Co
l
o
r
Imag
e
Co
m
p
res
s
i
o
n
Bas
e
d
o
n
W
av
e
l
et
,
D
i
fferen
t
i
a
l
Pu
l
s
e
C
o
d
e
Mo
d
u
l
at
i
o
n
an
d
Q
u
a
d
t
ree
Co
d
i
n
g
,
"
R
es
a
r
c
h
Jo
u
r
n
a
l
o
f
A
p
p
l
i
e
d
S
c
i
en
ce,
E
n
g
i
n
ee
r
i
n
g
a
n
d
Tech
n
o
l
o
g
y
,
v
o
l
.
1
4
,
n
o
.
2
,
p
p
.
7
3
-
7
9
,
2
0
1
7
.
[1
2
]
Y
.
Z
h
an
g
,
Y.
F.
Pu
,
J.
R.
H
u
an
d
J.
L
.
Z
h
o
u
,
"
A
C
l
as
s
o
f
Fract
i
o
n
al
-
O
rd
er
V
ari
a
t
i
o
n
a
l
Imag
e
I
n
p
a
i
n
t
i
n
g
Mo
d
e
l
s
,
"
A
p
p
l
.
M
a
t
h
.
In
f
.
S
c
i
.
,
v
o
l
.
6
,
n
o
.
2
,
p
p
.
2
9
9
-
3
0
6
,
2
0
1
2
.
[1
3
]
W
.
M.
A
b
d
-
E
l
h
af
i
ez
,
O
mar
Re
y
ad
,
M.
A
.
Mo
fa
d
d
e
l
,
M
o
h
ame
d
Fat
h
y
,
"
Imag
e
E
n
cry
p
t
i
o
n
A
l
g
o
ri
t
h
m
Me
t
h
o
d
o
l
o
g
y
Bas
ed
o
n
Mu
l
t
i
ma
p
p
i
n
g
Imag
e
Pi
x
el
,
"
Th
e
4
t
h
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
A
d
va
n
ce
d
M
a
c
h
i
n
e
Lea
r
n
i
n
g
Tech
n
o
l
o
g
i
es
a
n
d
A
p
p
l
i
ca
t
i
o
n
s
(A
M
LT
A
2
0
1
9
)
, v
ol
.
9
2
1
,
p
p
.
6
4
5
-
6
5
5
,
2
0
1
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
A
n
e
ff
icie
nt
c
olor
image
c
ompr
e
s
s
ion
tec
hnique
(
W
alaa
M
.
A
bd
-
E
lhaf
iez
)
2377
[1
4
]
Y
an
Fen
g
,
H
u
a
L
u
,
X
i
L
i
an
g
Z
en
g
,
"
A
Fract
a
l
Imag
e
Co
mp
res
s
i
o
n
Met
h
o
d
Bas
e
d
o
n
M
u
l
t
i
-
W
av
e
l
et
,
"
TE
LKO
M
NIKA
Tel
eco
m
m
u
n
i
ca
t
i
o
n
Co
m
p
u
t
i
n
g
El
ect
r
o
n
i
c
s
a
n
d
Co
n
t
r
o
l
,
v
o
l
.
1
3
,
n
o
.
3
,
p
p
.
9
9
6
-
1
0
0
5
,
2
0
1
5
.
[1
5
]
L
ei
Z
h
u
,
J
i
al
i
e
Sh
e
n
,
L
i
an
g
X
i
e,
Z
h
i
y
o
n
g
Ch
e
n
g
,
"
U
n
s
u
p
er
v
i
s
ed
V
i
s
u
al
H
as
h
i
n
g
w
i
t
h
Seman
t
i
c
A
s
s
i
s
t
an
t
fo
r
Co
n
t
en
t
-
Bas
e
d
Imag
e
Ret
ri
e
v
al
,
"
IE
E
E
Tr
a
n
s
a
c
t
i
o
n
s
o
n
Kn
o
wl
e
d
g
e
a
n
d
D
a
t
a
E
n
g
i
n
eer
i
n
g
,
v
o
l
.
29
,
n
o
.
2,
pp.
4
72
-
4
8
6
,
2
0
1
7
.
[1
6
]
L
i
an
g
X
i
e,
J
i
a
l
i
e
Sh
e
n
,
J
u
n
g
o
n
g
H
an
,
L
ei
Z
h
u
,
L
i
n
g
Sh
ao
,
"
D
y
n
am
i
c
M
u
l
t
i
-
V
i
e
w
H
a
s
h
i
n
g
f
o
r
O
n
l
i
n
e
Im
ag
e
Ret
ri
e
v
al
,
"
Twen
t
y
-
S
i
x
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
Jo
i
n
t
C
o
n
f
er
e
n
ce
o
n
A
r
t
i
f
i
ci
a
l
I
n
t
e
l
l
i
g
e
n
ce
,
p
p
.
3
1
3
3
-
3
1
3
9
,
2
0
1
7
.
[1
7
]
Man
d
y
D
o
u
g
l
a
s
,
K
are
n
Bai
l
ey
,
Mar
k
L
een
e
y
,
K
e
v
i
n
Cu
rran
,
"
U
s
i
n
g
S
V
D
a
n
d
D
W
T
Ba
s
ed
St
e
g
an
o
g
ra
p
h
y
t
o
E
n
h
an
ce
t
h
e
Secu
ri
t
y
o
f
W
at
ermar
k
ed
Fi
n
g
er
p
ri
n
t
I
mag
es
,
"
TE
LKO
M
NIK
A
Tel
ec
o
m
m
u
n
i
c
a
t
i
o
n
,
Co
m
p
u
t
i
n
g
,
E
l
ec
t
r
o
n
i
cs
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
1
5
,
n
o
.
3
,
p
p
.
1
3
6
8
-
1
3
7
9
,
2
0
1
7
.
[1
8
]
A
l
e
x
an
d
r
e
Z
ag
h
et
t
o
,
Ri
car
d
o
L
.
d
e
Q
u
ei
r
o
z
,
"
Scan
n
ed
D
o
c
u
men
t
Co
mp
re
s
s
i
o
n
U
s
i
n
g
Bl
o
ck
-
Ba
s
ed
H
y
b
ri
d
V
i
d
e
o
Co
d
ec,
"
IE
E
E
Tr
a
n
s
a
ct
i
o
n
s
o
n
Im
a
g
e
P
r
o
ces
s
i
n
g
(TI
P
)
,
v
o
l
.
2
2
,
n
o
.
6
,
p
p
.
2
4
2
0
-
2
4
2
8
,
2
0
1
3
.
[1
9
]
W
al
aa
M.
A
b
d
-
E
l
h
af
i
ez,
Mo
h
ame
d
H
es
h
mat
,
“Med
i
cal
I
mag
e
E
n
cry
p
t
i
o
n
V
i
a
L
i
ft
i
n
g
Met
h
o
d
,
”
Jo
u
r
n
a
l
o
f
In
t
el
l
i
g
en
t
&
F
u
z
z
y
S
y
s
t
e
m
s
,
v
o
l
.
3
8
,
n
o
.
3
,
p
p
.
2
8
2
3
-
2
8
3
2
,
2
0
2
0
.
[2
0
]
Q
i
a
n
g
Z
h
an
g
,
an
d
X
i
ao
p
en
g
W
e
i
,
"
A
n
E
ff
i
ci
e
n
t
A
p
p
ro
a
ch
fo
r
D
N
A
Frac
t
al
-
b
as
e
d
Imag
e
E
n
cr
y
p
t
i
o
n
,
"
A
p
p
l
.
M
a
t
h
.
In
f
.
S
c
i
.
,
v
o
l
.
5
,
n
o
.
3
,
p
p
.
4
4
5
-
4
5
9
,
2
0
1
1
.
[2
1
]
W
an
g
X
u
e
-
g
u
an
g
,
Ch
en
Sh
u
-
h
o
n
g
,
"
A
n
Imp
r
o
v
e
d
I
mag
e
Seg
men
t
at
i
o
n
A
l
g
o
r
i
t
h
m
Bas
ed
o
n
T
w
o
-
D
i
men
s
i
o
n
a
l
O
t
s
u
Met
h
o
d
,
"
In
f
.
S
c
i
.
Let
t
.
,
v
o
l
.
1
,
n
o
.
2
,
p
p
.
77
-
8
3
,
2
0
1
2
.
[2
2
]
A
.
Sh
i
o
,
"
A
n
A
u
t
o
ma
t
i
c
T
h
res
h
o
l
d
i
n
g
A
l
g
o
ri
t
h
m
Bas
e
d
O
n
A
n
I
l
l
u
mi
n
at
i
o
n
-
In
d
ep
e
n
d
e
n
t
Co
n
t
ra
s
t
Mea
s
u
re,
"
IE
E
E
Co
m
p
u
t
er
S
o
ci
e
t
y
Co
n
f
e
r
en
ce
o
n
Co
m
p
u
t
e
r
V
i
s
i
o
n
a
n
d
P
a
t
t
e
r
n
R
ec
o
g
n
i
t
i
o
n
,
p
p
.
6
3
2
-
6
3
7
,
1
9
8
9
.
[2
3
]
Fai
s
a
l
Sh
afai
t
,
D
a
n
i
e
l
K
ey
s
ers
,
T
h
o
mas
M.
Breu
el
,
"
E
ffi
c
i
en
t
Imp
l
eme
n
t
a
t
i
o
n
o
f
L
o
cal
A
d
a
p
t
i
v
e
T
h
res
h
o
l
d
i
n
g
T
ech
n
i
q
u
e
s
U
s
i
n
g
In
t
eg
ra
l
Imag
es
,
"
S
P
IE
D
o
cu
m
e
n
t
R
e
co
g
n
i
t
i
o
n
a
n
d
R
et
r
i
ev
a
l
X
V
,
2
0
0
8
.
[2
4
]
W.
M.
Abd
-
E
l
h
afi
ez,
U
.
S.
Mo
h
ammed
,
A
d
em
K
,
"
O
n
H
i
g
h
Perfo
rma
n
ce
Imag
e
Co
mp
res
s
i
o
n
T
ec
h
n
i
q
u
e,
"
S
ci
e
n
ceA
s
i
a
,
v
o
l
.
3
9
,
p
p
.
4
1
6
-
4
2
2
,
2
0
1
3
.
[2
5
]
J
.
F
.
Can
n
y
,
"
A
Co
m
p
u
t
at
i
o
n
al
A
p
p
ro
ac
h
t
o
E
d
g
e
D
e
t
ec
t
i
o
n
,
"
IE
E
E
T
r
a
n
s
a
ct
i
o
n
s
o
n
P
a
t
t
e
r
n
A
n
a
l
y
s
i
s
a
n
d
M
a
c
h
i
n
e
In
t
e
l
l
i
g
e
n
ce
,
v
o
l
.
8
,
n
o
.
6
,
p
p
.
6
7
9
-
6
9
8
,
1
9
8
6
.
[2
6
]
W
.
M.
A
b
d
-
E
l
h
afi
ez,
W
aj
e
b
G
h
ari
b
i
,
"
Co
l
o
r
Imag
e
C
o
mp
res
s
i
o
n
A
l
g
o
ri
t
h
m
Ba
s
ed
o
n
D
CT
B
l
o
c
k
s
,
"
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
Co
m
p
u
t
e
r
S
ci
e
n
ce
,
v
o
l
.
9
,
n
o
.
4
,
p
p
.
3
2
3
-
3
2
8
,
2
0
1
2
.
B
I
OG
RA
P
HI
E
S
OF
AU
T
HO
RS
W
a
l
a
a
M
.
A
b
d
-
E
l
h
a
f
i
e
z
recei
v
ed
h
er
B.
Sc.
an
d
M.
Sc.
d
eg
ree
s
fro
m
s
o
u
t
h
v
a
l
l
e
y
u
n
i
v
er
s
i
t
y
,
So
h
ag
b
ra
n
ch
,
So
h
a
g
,
E
g
y
p
t
i
n
2
0
0
2
an
d
fro
m
So
h
a
g
U
n
i
v
er
s
i
t
y
,
So
h
a
g
,
E
g
y
p
t
,
J
an
2
0
0
7
,
res
p
ec
t
i
v
el
y
,
an
d
h
er
Ph
.
D
.
d
eg
ree
fro
m
S
o
h
a
g
U
n
i
v
er
s
i
t
y
,
So
h
a
g
,
E
g
y
p
t
.
H
er
re
s
earch
i
n
t
ere
s
t
s
i
n
c
l
u
d
e
i
mag
e
s
eg
me
n
t
a
t
i
o
n
,
i
mag
e
en
h
an
cem
en
t
,
i
mag
e
reco
g
n
i
t
i
o
n
,
i
mag
e
co
d
i
n
g
,
v
i
d
eo
co
d
i
n
g
,
an
d
t
h
ei
r
a
p
p
l
i
ca
t
i
o
n
s
i
n
i
mag
e
p
r
o
ces
s
i
n
g
.
W
a
j
eb
G
h
a
r
i
b
i
,
Pro
fes
s
o
r
o
f
Co
mp
u
t
er
Sci
e
n
ce.
H
e
g
o
t
h
i
s
Ph
.
D
fro
m
Bel
aru
s
A
cad
emy
o
f
Sci
en
ce
s
i
n
1
9
9
0
.
H
i
s
res
earch
i
n
t
ere
s
t
s
i
n
cl
u
d
e
Cy
b
ers
ecu
ri
t
y
,
Mach
i
n
e
L
earn
i
n
g
,
So
ft
w
are
E
n
g
i
n
eeri
n
g
,
Q
u
an
t
u
m
C
o
mp
u
t
i
n
g
an
d
O
p
t
i
m
i
zat
i
o
n
.
H
e
h
as
mo
re
t
h
a
n
1
3
5
p
u
b
l
i
s
h
ed
re
s
earch
p
ap
er
s
i
n
re
p
u
t
ed
j
o
u
r
n
al
s
an
d
co
n
fe
ren
ce
s
.
M
o
ha
m
ed
Hes
h
m
a
t
,
recei
v
e
d
h
i
s
B.
Sc.
A
n
d
M.
Sc.
D
eg
rees
fro
m
So
u
t
h
V
al
l
ey
U
n
i
v
ers
i
t
y
,
So
h
ag
b
ran
c
h
,
So
h
ag
,
E
g
y
p
t
i
n
2
0
0
2
an
d
h
i
s
P
h
.
D
.
d
e
g
ree
fro
m
So
h
ag
U
n
i
v
er
s
i
t
y
,
So
h
ag
,
E
g
y
p
t
an
d
Bau
h
au
s
-
U
n
i
v
ers
i
t
y
,
W
e
i
mar,
G
erman
y
,
2
0
1
0
.
H
i
s
res
ea
rch
i
n
t
eres
t
s
i
n
c
l
u
d
e
co
mp
u
t
er
v
i
s
i
o
n
,
3
D
d
a
t
a
acq
u
i
s
i
t
i
o
n
,
o
b
j
ect
rec
o
n
s
t
ru
ct
i
o
n
,
i
ma
g
e
s
e
g
men
t
at
i
o
n
,
i
mag
e
en
h
a
n
cemen
t
an
d
i
m
ag
e
rec
o
g
n
i
t
i
o
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.