Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
11
,
No.
3
,
June
2021
,
pp. 1
985~
1993
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v11
i
3
.
pp1985
-
19
93
1985
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om
Novel
h
yb
rid
f
ra
mework
for
i
ma
ge
c
omp
ressi
on fo
r
s
up
po
rtive
h
ar
dw
are
d
esign
of
b
oo
s
ti
ng
c
omp
ression
Prem
a
ch
and
D
.
R
.
,
U.
Er
anna
Depa
rtment
o
f
E
le
c
troni
cs
and
C
om
m
unic
at
ion
E
ngine
er
ing
,
Ball
a
ri
Inst
it
ut
e
of
T
ec
hnolog
y
and
Mana
gement
,
B
al
l
ari
,
India
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
N
ov
10
, 201
9
Re
vised
Oct
1
2
,
20
20
Accepte
d
Nov 1
, 2
0
20
Perform
ing
the
i
m
age
compress
ion
over
the
reso
urc
e
constraine
d
har
dwar
e
is
quit
e
a
cha
l
le
ngi
ng
ta
sk.
Althoug
h,
the
r
e
has
been
var
ious
appr
oa
che
s
bei
n
g
ca
rri
ed
out
towa
rds
image
compr
ession
conside
ri
ng
the
har
dware
aspe
ct
of
it,
but
stil
l
th
ere
are
prob
le
m
s
a
ss
oci
at
ed
with
the
m
emor
y
acce
l
era
t
ion
associa
t
ed
with
the
en
ti
re
oper
ation
that
downgrade
th
e
per
fo
rm
anc
e
of
th
e
har
dware
d
evi
c
e
.
The
r
efo
re
,
th
e
proposed
appr
oa
ch
pre
sents
a
co
st
eff
ective
image
compress
ion
m
ec
hani
sm
which
offe
rs
lossless
compres
sion
us
ing
a
unique
combin
at
ion
of
the
n
on
-
li
ne
ar
fil
t
ering,
segm
ent
atio
n,
cont
our
det
e
ct
ion
,
fo
ll
ow
ed
b
y
the
opti
m
i
za
t
ion.
The
com
pre
ss
ion
m
ec
han
ism
ada
pts
ana
l
y
t
ical
appr
o
ac
h
for
signifi
c
a
nt
image
compress
ion.
The
exec
uti
on
of
the
compress
ion
m
ec
hani
sm
y
ie
lds
faste
r
r
esponse
ti
m
e
,
r
educ
ed
m
ea
n
squar
e
err
or, improved s
igna
l
qu
al
i
t
y
an
d
significant c
o
m
pre
ss
ion
rat
io
per
form
anc
e
.
Ke
yw
or
d
s
:
Enc
od
i
ng
Hardwa
re
a
cce
le
rati
on
Im
age
c
om
pr
ession
Segm
entat
ion
VLSI
a
rch
it
ect
ur
e
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Pr
em
a
chan
d
D
.
R
.
Dep
a
rtm
ent o
f
Ele
ct
ro
nics
and C
omm
un
ic
ation
En
gin
ee
rin
g
Ba
ll
ari In
sti
tut
e of Tec
hnolog
y and M
ana
ge
m
ent
Ba
ll
ari,
India
Em
a
il
:
pr
em
ch
andbitm
@g
m
a
il
.co
m
1.
INTROD
U
CTION
Perfo
rm
ing
dif
fer
e
nt
f
or
m
s
of
pr
ocessin
g
over
a
com
plex
form
of
a
n
im
age
is
a
chal
le
ng
in
g
ta
s
k
wh
e
n
it
com
es
to
hard
war
e
-
ba
sed
im
ple
m
ent
at
ion
.
I
n
this
a
sp
ect
,
im
age
c
om
pr
essio
n
is
on
e
s
uc
h
proce
ss
tha
t
sign
ific
a
ntly
offer
s
over
hea
d
i
n
the
ha
rdwa
re
reali
zat
ion
.
Al
l
the
pro
blem
s
sta
rts
with
t
he
m
e
m
or
y
syst
em
of
the
ha
r
dw
a
re
c
om
po
ne
nt
w
hile
proces
sin
g
la
rg
e
num
ber
of
i
m
age
pix
el
s.
I
t
is
belie
ve
d
th
at
the
pe
rfo
rm
anc
e
of
the
m
e
m
or
y
m
anag
em
ent
so
le
ly
de
pends
upon
t
he
patte
rn
of
it
s
acce
ss
ing
m
echan
is
m
.
It
is
esse
ntial
tha
t
analy
sis
of
th
e
traces
of
th
e
m
e
m
or
y
is
require
d
to
be
inv
e
sti
gated
in
or
der
t
o
i
m
pr
ov
e
t
he
m
e
m
ory
m
anag
em
ent
of
the
ha
r
dw
a
re
.
T
he
im
age
com
pr
essio
n
m
echan
ism
aim
s
to
op
ti
m
iz
e
t
he
siz
e
of
im
a
ge
for
sm
oo
th
transm
issi
on
a
nd
stora
ge
[
1].
T
he
pr
esen
ce
of
a
rtif
act
s
in
the
im
a
ges
ca
n
al
so
re
m
ov
e
throu
gh
i
m
age
com
pr
essio
n
[
2].
But,
the
re
i
s
chall
en
ging
sit
uation
i
n
ac
hiv
in
g
sig
nific
ant
im
age
com
pr
essi
on
f
or
th
e
are
a
and ene
gy c
onstrai
ned d
e
vice
s [3].
In
or
der
to
pe
r
form
an
eff
ic
ien
t
i
m
age
co
m
pr
essio
n,
the
re
are
var
i
ou
s
c
onditi
on
s
that
are
req
ui
red
t
o
be
cat
ere
d
up.
The
pr
im
ary
c
onditi
ons
to
be
sat
isfie
d
are
t
hat
the
usa
ge
of
lossless
c
ompressi
on
is
i
m
portan
t
for
resist
in
g
l
os
s
of
trace
d
data
in
m
e
m
or
y
syst
e
m
.
The
seconda
ry
co
nd
it
io
n
to
be
sat
isfie
d
is
tha
t
there
sh
oul
d
be
hi
gher
de
gr
ee
of
thr
oughput
an
d
inc
rease
d
s
pe
ed
f
or
hi
gh
e
r
ba
ndwidt
h
s
uport
abilt
y.
W
it
h
the
increasin
g
fr
e
quency
of
op
e
ra
ti
on
a
nd
the
hi
gh
e
r
data
band
width,
the
re
is
a
nee
d
of
hi
gher
r
at
e
of
th
rou
ghput
too
.
T
he
m
a
in
reason
be
hind
this
is
that
in
o
rd
e
r
to
offer
be
tt
er
analy
sis
of
m
e
m
or
y
trac
es
there
is
a
need
of
dev
el
op
i
ng
acc
el
erati
on
syst
e
m
of
the
com
pressi
on
with
re
sp
ect
to
ha
rdw
are
m
od
el
ing
a
nd
not
the
soft
war
e
-
base
d
sym
pto
m
at
ic
app
r
oac
h.
A
no
t
her
im
portant
fin
ding
is
t
hat
s
of
t
war
e
-
ba
sed
co
m
pr
ession
a
pp
ro
ac
h
is
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021
:
19
85
-
1993
1986
al
ways
slow
e
r
in
com
par
iso
n
to
the
ha
rdware
-
base
d
c
om
pr
ession
a
ppr
oac
h
that
dow
ngra
des
the
c
om
pr
essio
n
perform
ance.
Ther
e
f
or
e,
the
r
e
is
a
sign
ific
ant
trade
off
bet
ween
hardware
-
b
ase
d
ap
proac
h
an
d
software
-
base
d
appr
oach
i
n
or
der
t
o
achie
ve
the
ta
rg
et
of
com
pr
essio
n
e
ff
ic
ie
ncy.
T
he
increase
d
use
of
In
te
rn
et
-
of
-
Things
(IoT)
ca
n
le
ad
to
m
or
e
us
a
ge
of
re
source
c
onstrai
nt
de
vice
s
connecte
d
ea
ch
oth
e
r
th
rou
gh
diff
e
re
nt
ne
tworks
[4
]
. T
he
im
age
co
m
pr
essio
n
c
an be ac
hiev
d
t
hro
ugh
s
of
t
ware/
hardw
a
re
bas
ed
a
ppr
oach
es
[5
-
7].
Most
of
t
he
software
a
ppr
oac
hes
are
de
pend
ed
on
vector
m
ulti
plica
ti
on
an
d
ot
her
m
at
rix
op
e
rati
ons
[8
]
.
But,
these o
pe
rati
ons
are n
ot
feasi
ble
for
the
resour
ce
c
on
st
raine
d
de
vi
ces
du
e
to
inc
r
eased
com
pu
ta
ti
on
al
com
plexity
.
T
he
existi
ng
m
echan
ism
s
are
no
t
m
uch
effi
ci
ent
to
han
dl
e
these
pr
obl
e
m
s
and
offer
bette
r
su
pp
or
t
to
pro
vid
e
hardw
a
re
eff
ic
ie
nt
so
l
ution
for
com
pr
e
ssion
te
c
hn
i
ques
[9
,
10]
.
He
nc
e,
this
pa
per
offer
s
a
cost
e
ff
ect
ive
so
luti
on
pro
vid
in
g
bala
nc
e
between
si
gn
al
qual
it
y
and
c
om
pr
ession
thr
ough
hybri
d
com
pr
essio
n
te
chn
i
qu
e
in
wi
r
e
le
ss
sens
or
ne
twork
(
WSN)
.
The
s
ect
io
n
2
of
the
pap
e
r
e
xp
la
in
s
the
e
xisti
ng
researc
h
a
naly
sis
an
d
pro
blem
descr
ipti
on
in
s
ect
io
n
3.
The
p
rop
os
ed
te
chn
iq
ue
al
on
g
with
t
he
al
gorithm
i
m
ple
m
entat
io
n
is
el
ab
or
at
e
d
in
s
ect
ion
5.
The
re
su
lt
s
dis
cussion
with
c
om
par
at
ive
an
al
ysi
s
is
hig
hligh
te
d
i
n
sect
ion
6
a
nd c
on
cl
us
io
n
i
n
s
e
ct
ion
7.
The
st
ud
y
of
va
rio
us
e
xisti
ng
com
pr
essio
n
t
echn
i
qu
e
s
f
or
dig
it
al
i
m
age
is
disc
us
se
d
in
this
sect
ion.
In
Ca
bro
nero
et
al
.
[11],
a
Hybr
i
d
c
om
pr
ession
ap
proac
h
is
a
da
pted
f
or
tran
sf
or
m
at
i
on
-
op
e
rati
on
is
carrie
d
ou
t
ove
r
colo
r
ed
m
os
ai
c
i
m
a
ge.
A
s
ubj
ect
i
ve
optim
iz
at
io
n
process
with
su
pe
rv
ise
d
le
arn
i
ng
al
gorith
m
s
is
introd
uc
ed
in
Am
irj
anov
a
nd
Dim
i
li
le
r
[1
2]
over
m
edical
im
age.
I
n
Wang
et
al
.
[
13]
,
a
li
ne
-
buf
fer
sc
hem
e
is
introd
uced
for
com
pr
essio
n
over
the
proces
sing
ci
rc
uits.
A
w
ork
of
Ch
en
et
al
.
[
14
]
pro
vid
e
d
a
color
filt
e
r
arr
ay
al
ong
with
an
im
pr
ov
e
d
G
olo
m
b
-
Ri
ce
cod
e
to
ac
hi
eve
sig
nificant
on
-
chi
p
com
pr
essio
n.
A
ha
r
dw
a
r
e
acce
le
rati
on
a
ppr
oac
h
f
or
inte
gr
at
e
d
ci
rc
uits
com
pr
essio
n
i
s
intr
oduce
d
i
n
Halawa
ni
et
al
.
[15].
C
ho
i
et
al
.
[16]
ha
ve
c
onsidere
d
hardware
th
rou
ghput
appr
oach
f
or
on
-
c
hip
c
om
pr
ession.
A
com
pr
essi
on
sch
e
m
e
for
i
m
age
sens
or
s
is
pr
ese
nte
d
in
Ka
ur
et
al
.
[
17
]
by
us
i
ng
i
nter
po
la
ti
on
operati
on
ov
e
r
t
he
on
-
chi
p
hardw
a
r
e
desig
n.
I
n
O
ni
sh
i
et
al
.
[18],
a
m
ultim
edia
encoder
a
ppro
ac
h
is
introdu
ce
d
ove
r
sing
le
-
chi
p
desi
gn
a
nd
achieve
d
scal
a
ble
com
pr
essi
on
pro
cess.
T
he
im
ple
m
enta
ti
on
of
tran
sf
orm
-
based
c
ompressi
on
sc
he
m
e
fo
r
com
pr
essio
n
is
d
esc
ribe
d
in
Z
hu
et
a
l
.
[1
9] by
u
sin
g bloc
k
c
od
i
ng.
A
V
LSI
arc
hitec
ture
with
tr
ee
-
base
d
a
ppr
oach
f
or
t
ow
a
rd
s
im
age
enco
de
r
base
d
co
m
pr
ession
i
s
giv
e
n
in
H
sie
h
et
al
.
[20].
An
ener
gy
ef
fici
en
t
i
m
ple
m
entat
i
on
of
VLSI
a
rc
hitec
ture
f
or
i
m
age
com
pr
es
sion
is
introd
uced
in
Chen
et
al
.
[21]
con
si
der
i
ng
ultra
-
high
de
f
init
ion
f
ram
e.
Gr
a
dient
-
base
d
appro
ac
h
to
w
ards
i
m
pr
ovin
g
co
m
pr
ession
pe
rfor
m
ance
is
seen
in
the
w
or
k
of
Kim
et
al
.
[22].
Luca
s
et
al
.
[23]
have
us
e
d
pr
e
dicti
ve
-
base
d
sc
hem
e
fo
r
t
he
purpose
of
perform
ing
los
sle
ss
com
pr
ess
ion
sc
hem
e.
The
work
car
ried
out
by
the
Yi
n
et
al
.
[24]
hav
e
us
e
d
a
n
inte
gral
histo
gr
am
ba
sed
a
ppr
oac
h
to
acce
le
rate
t
he
m
echan
ism
of
the
featur
e
ext
racti
on
proc
es
s
f
or
co
ntr
olli
ng
bette
r
c
om
pu
ta
ti
on
al
oper
at
ion
ov
e
r
t
he
m
e
m
or
y
ac
cess
i
n
hard
war
e.
Pa
rikh
et
al
.
[
25
]
hav
e
use
d
la
te
st
enc
od
e
rs
for
pe
rfor
m
ing
m
axim
iz
ed
com
pr
essi
on
pe
r
for
m
ance
for
the
com
plex
f
or
m
of
an
i
m
age.
The
stud
y
has
use
d
the
pa
rtit
ion
i
ng
proce
ss
fo
r
bette
r
m
e
m
or
y
m
anag
em
ent
du
ri
ng
c
om
pr
es
sion
ov
e
r
par
al
le
l
arch
it
ect
ur
e
.
A
doptio
n
of
colo
r
filt
er
ar
ra
y
was
al
so
pr
oven
t
o
be
bette
r
s
olut
ion
f
or
perf
orm
ing
com
pr
ession
es
pecial
ly
in
the
case
of
a
pp
li
cat
io
n
that
dem
and
s
li
ve
com
pr
essio
n
operati
on
t
o
be
ta
kin
g
place.
Wor
k
in
t
his
di
recti
on
has
be
en
car
ried
out
by
Che
n
et
al
.
[26]
wh
e
re
G
olo
m
b
Ri
ce
is
c
on
sidere
d
for
i
m
ple
m
entat
ion
f
or
en
ha
ncin
g
th
e
c
om
pr
ession
ef
fici
enc
y.
Th
e
pro
blem
s
associat
ed
with
c
olo
r
c
om
pr
essio
n
is
car
ried
ou
t
by
Pete
r
et
al
.
[27]
w
her
e
the
aut
hors
ha
ve
us
e
d
diffusi
on
-
ba
se
d
ap
proac
h
al
ong
with
i
np
ai
nting
phen
om
e
non
f
or
offer
i
ng
s
uperio
r
qu
al
it
y
of
reco
ns
tructed
i
m
age.
Zha
ng
et
al
.
[
28
]
w
her
e
a
de
finiti
ve
trans
f
or
m
at
ion
sche
m
e
has
been
presente
d.
T
his
appr
oach
ha
s
been
pr
ese
nted
fo
r
the
up
gr
a
ding
the
existi
ng
JP
EG
c
ompressi
on
perfor
m
ance.
Ther
e
f
or
e
,
there
are
var
i
ous
appr
oach
es
t
owar
ds
im
age
com
pr
essio
n
w
he
re
ha
rdwa
re
appr
oach
has
be
en
co
ns
id
ere
d.
Stud
y
f
oc
us
in
g
over
the
energy
eff
i
ci
ency
as
well
as
con
strai
nt
area
of
t
he
ci
ru
it
de
sig
n
co
nnect
ing
with
the
co
ncep
t
of
i
m
age
com
pr
essio
n
ha
s
bee
n
disc
usse
d
by
Zei
nola
bed
in
et
al
.
[
29
]
.
Kim
et
al
.
[
30
]
has
us
e
d
e
ncodin
g
-
base
d
appr
oach
f
or
c
arr
yi
ng
the
co
m
pr
ession
op
e
rati
on
w
her
e
the
pr
el
im
inary
enc
od
i
ng
co
m
pu
te
s
the
bitst
ream
le
ng
th
w
hile
the
sec
ondar
y
e
ncode
r
is
res
pons
i
ble
f
or
ge
ner
at
io
n
of
st
r
ea
m
s
of
the
da
ta
pack
et
s
ov
e
r
VL
SI
arch
it
ect
ure.
St
ud
y
c
onsideri
ng
the
VLSI
arc
hitec
ture
to
wa
r
ds
the
c
om
pr
es
sion
a
ppr
oach
has
been
ca
rr
ie
d
out
by the
ne
xt sec
ti
on
ou
tl
ines
th
e resea
rch issu
es.
The
ov
e
r
view
of
e
xisti
ng
stu
dies
s
uggests
t
hat
the
c
om
pr
e
ssion
schem
es
are
fail
ed
to
a
dopt
t
he
VL
S
I
i
m
ple
m
entat
io
n
e
ff
ect
ively
.
S
om
e o
f
the
r
ese
arch i
ssu
es
are:
-
Lack
of h
a
r
dware effi
ci
ent sc
hem
es to mi
nim
iz
e com
pu
ta
t
ion
al
c
om
plexity
-
Least
co
ns
ide
r
at
ion
of f
il
t
erin
g process
in
l
ow
-
qual
it
y im
ag
e com
pr
essio
n
-
Com
plete
selection
of sig
nal pro
ces
sin
g
is
not consi
der
e
d
i
n pr
e
vio
i
us
wor
ks
-
I
ns
i
gn
i
ficant se
le
ct
ion
of c
ompressi
on f
eat
ures to yi
el
d
c
os
t
eff
ect
iv
eness
s
olu
ti
on
The
stu
dy
devel
op
s
the
c
ompu
ta
ti
onal
m
od
el
to
a
ddress
the
above
sta
te
d
researc
h
pro
blem
s.
The
flo
w
of
the
pr
opose
d
c
om
pr
ession
a
ppr
oach
is
il
lustrate
d
in
F
igure
1.
The
pro
po
se
d
syst
e
m
adap
ts
analy
ti
cal
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Novel
h
y
br
id
fr
am
ew
or
k f
or
i
mage
c
omp
res
sion f
or
sup
po
r
ti
ve ha
r
dwar
e
desig
n of
..
.
(
P
remach
and D
.
R
.
)
1987
appr
oach
wher
e
input
i
m
age
is
processe
d
f
or
cost
eff
ect
iv
e
i
m
age
com
pr
ession.
The
si
gn
i
ficant
aspect
of
this
so
luti
on
is
that
it
fo
ll
ow
s
hy
br
i
d
com
pr
ession
te
ch
nique
of
lossless
a
nd
lossy
com
pr
ession
ov
e
r
se
le
ct
ed
i
m
age sector a
nd co
m
pr
essi
on
ov
e
r regi
ons
of the im
age r
e
sp
ect
ively
.
The
pr
opos
e
d
syst
e
m
m
akes
us
e
of
the
sim
plifie
d
m
od
el
ing
t
hat
init
ia
tes
by
perf
or
m
i
ng
filt
erin
g
op
e
rati
on.
Thi
s
op
e
rati
on
as
sist
s
in
el
i
m
in
at
ing
possibil
it
y
of
any
furth
er
arti
facts
that
cou
ld
be
pos
sibly
pr
ese
nt
du
rin
g
the
tran
sm
issio
n
proces
s
of
the
im
ag
e
in
the
wi
reless
co
m
m
un
ic
at
ion
m
edium
.
The
filt
ered
i
m
age
is
subj
e
ct
ed
to
ne
xt
pr
ocess
w
her
e
uniq
uely
not
th
e
com
plete
i
m
age
is
s
ubj
ect
e
d
to
the
c
om
pr
ession.
The
c
om
plete
i
m
age
is
hypot
hetic
al
ly
cl
assif
ie
d
i
nto
tw
o
pa
rts
w
her
e
one
sect
or
of
the
i
m
age
is
c
onsid
ered
t
o
be
c
riti
cal
ly
i
m
po
rtant
w
he
r
eas
ot
her
pa
rt
of
the
im
age
is
not
that
im
po
rta
nt.
F
ro
m
app
li
cat
io
n
as
well
as
com
m
un
ic
at
ion
view
point,
it
is
essenti
al
that
i
m
po
rtant
sect
or
the
im
age
is
su
bject
ed
f
or
lossless
com
pr
e
ssion
schem
e
wh
il
e
the
not
s
o
i
m
portant
par
t
of
an
im
age
is
su
bject
e
d
to
the
lo
ssy
com
pr
essio
n
sc
hem
e.
T
he
pro
po
se
d
sc
he
m
e
us
es
co
nt
ours
detect
ions
an
d
op
ti
m
i
zat
ion
process
.
A
nother
ess
entia
l
par
t
of
the
i
m
ple
m
entat
io
n
of
the
pro
po
sed
syst
em
is
i
ts
segm
entat
io
n
proces
s
an
d
the
co
ntour
d
e
te
ct
ion
process
.
Thi
s
process
c
ontri
bu
te
s
t
ow
a
rds
the
extracti
on
of
t
he
feat
ur
es
that
is
us
e
d
f
or
m
ini
m
izing
the
dim
ension
s
of
a
n
i
m
age.
This
ch
arecte
risti
cs
of
the
con
to
urs
are
furthe
r
opti
m
iz
ed
wh
ere
c
on
cat
e
natio
n
operati
on
is
carr
ie
d
ou
t
ov
e
r
th
e
ext
ra
ct
ed
re
d
gr
ee
n
an
d
blu
e
c
ompone
nts
of
a
n
i
m
age
wh
ic
h
m
ai
ntains
si
m
i
la
r
dim
ension
s
of
a
n
i
m
age
bu
t
with
m
or
e
eff
ect
ive
inf
orm
ation
of
the
c
on
t
ours.
T
he
finall
y
ob
ta
ine
d
opti
m
iz
ed
con
to
ur
s
ar
e
su
bject
e
d
f
or
t
he
wav
el
et
c
om
pr
ession
t
his
re
nd
e
rs
t
he
process
of
com
pressi
on
acce
le
r
at
ion
s
wifter
over
the
hard
war
e.
T
he next
sect
ion di
scussion
of the
all
the esse
ntial
p
r
ocess
b
l
ocks sho
wn in
F
ig
ur
e
1.
I
n
p
u
t
I
m
a
g
e
P
r
o
c
e
s
s
i
n
g
I
n
p
u
t
I
m
a
g
e
P
e
r
f
o
r
m
i
n
g
F
i
l
t
e
r
i
n
g
O
p
e
r
a
t
i
o
n
S
e
c
t
o
r
-
S
e
l
e
c
t
i
v
e
C
o
m
p
r
e
s
s
i
o
n
S
e
g
m
e
n
t
a
t
i
o
n
-
O
p
e
r
a
t
i
o
n
C
o
n
t
o
u
r
D
e
t
e
c
t
i
o
n
A
p
p
l
y
i
n
g
O
p
t
i
m
i
z
e
d
C
o
n
t
o
u
r
s
D
e
t
e
c
t
i
o
n
W
a
v
e
l
e
t
-
b
a
s
e
d
C
o
m
p
r
e
s
s
i
o
n
C
o
m
p
r
e
s
s
e
d
I
m
a
g
e
Figure
1
.
S
yst
em
a
rch
it
ect
ur
e
2.
DESIG
N
P
RI
NC
I
PLE
OF
I
MPLEME
NT
ATIO
N
Desig
n
play
s
a
sign
ific
a
nt
r
ole
in
de
velo
ping
arc
hitec
ture
.
It
shou
l
d
be
no
te
d
that
an
e
ff
e
ct
ive
desi
gn
for
VLSI
a
rch
i
te
ct
ur
e d
em
ands
a
s
uitable
c
om
pu
ta
ti
on
al
m
od
el
t
hat p
er
form
s
m
anag
em
e
nt
an
d
str
uctur
i
zat
ion
of
al
l
the
proc
e
sses
in
e
ff
ic
ie
nt
m
ann
er
.
T
his
sect
ion
br
ie
fs
s
trat
egies
f
ollo
wed
in
im
age
com
pr
essio
n
w
he
re
it
m
ai
ntains
reconstr
ucted
im
age
qu
al
it
y
and
com
pu
ta
ti
on
al
com
plexity
.
The
qual
it
y
i
m
a
ge
is
co
ns
ide
r
ed
to
transm
it
in
the
pr
im
ar
y
as
s
umptio
n
a
nd
it
is
cha
racteri
zed
with
a
rtifact
s.
The
el
im
inat
i
on
of
the
artef
act
s
ar
e
consi
der
e
d
sec
ondary
as
s
ump
ti
on
by
up
gr
a
di
ng
ne
w
c
om
pressi
on
pr
ocess
.
The
lossless
c
om
pr
essio
n
is
aim
ed
in
te
rti
ar
y
as
s
umptio
n
over
spe
ci
fic re
gion
of i
m
age
.
2
.
1
.
Str
ateg
y fo
r
Implem
en
tation
The
pro
pose
d
syst
e
m
plann
ed
stragical
ly
to
pr
ovid
e
the
co
st
eff
ect
ive
i
m
age
com
pr
essi
on
a
nd
ob
ta
i
n
the
qual
it
y
i
m
age
afte
r
dec
om
pr
ession.
T
o
achive
t
his,
s
om
e
par
t
of
t
he
i
m
age
are
co
ns
ide
red
t
o
sel
ect
the
sect
or
of
t
he
i
m
age
for
c
om
pr
essio
n
.
T
he
st
rategy
al
so
off
ers
bette
r
ap
proach
f
or
hard
w
are
vie
wpoint
,
wh
e
re
the
m
e
m
or
y
allocati
on
s
ov
e
r
the
hard
war
e
im
age
com
pr
ession
is
co
ns
ide
red
that
can
re
du
ce
the
c
om
plexity
.
The
al
gorithm
ab
le
to p
rovide
the
c
om
pr
esse
d
im
age
with
f
ollow
i
ng steps
:
Algori
th
m
of
secto
r
-
sel
ecti
ve
co
m
pressio
n
appro
ach
Input
: Input Image (I)
Output
: Compressed image (ϕ)
Start
Step
-
1
:
init
ialize
I
Step
-
2
:
K
f
1
(I)
Step
-
3
:
I
4
f
2
(I)
Step
-
4
:
BW
f
3
(I
4
)
Step
-
5
:
R
co
nc
at
inate
(r
1
g
1
b
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021
:
19
85
-
1993
1988
Step
-
6
:
C
r
c
on
ca
t
inate
(r g b)
Step
-
7
:
C
p
c
on
ca
t
inate
(R G B)
Step
-
8
:
ϕ
f
4
(Cp)
End
The
al
gorithm
is ex
pr
esse
d
a
s
b
el
ow
:
a.
Inp
ut
i
m
age
processin
g
:
The
m
at
rix
(I
)
is
ge
ner
at
e
d
after
processi
ng
im
age
f
or
dig
it
iz
at
ion
(
Ste
p
-
1)
by
conve
rting R
G
B t
o gr
ey
scal
e
i
m
age (
im
age p
r
ocessi
ng stag
es are ill
us
te
d
i
n
Fi
gure
2
.
I
n
p
u
t
I
m
a
g
e
D
i
g
i
t
i
z
a
t
i
o
n
C
o
n
v
e
r
s
i
o
n
t
o
G
r
a
y
s
c
a
l
e
G
r
a
y
s
c
a
l
e
I
m
a
g
e
Figure
2
.
Stage
s of
i
np
ut
i
m
ag
e
p
ro
ces
sin
g
b.
Fil
te
ring
p
r
oc
ess
:
This
proc
ess
is
app
li
ed
to
el
i
m
inate
s
the
no
ise
in
the
i
m
age
befor
e
s
ubj
ect
in
g
to
com
pr
essio
n.
A
f
un
ct
i
on
f
1
(
x)
is
int
rod
uce
d
to
perform
non
-
li
near
filt
er
ing
that
al
s
o
im
pr
ov
es
the
i
m
age
processi
ng
sta
ge
(
Ste
p
-
2)
(
Non
li
nea
r
filt
erin
g
sta
ges
a
re
giv
en
in Fi
gure
3)
.
G
r
a
y
s
c
a
l
e
I
m
a
g
e
N
o
n
-
l
i
n
e
a
r
F
i
l
t
e
r
i
n
g
F
i
l
t
e
r
e
d
I
m
a
g
e
Figure
3
.
Stage
s of
n
on
-
l
i
near
f
il
te
ring
c.
Sect
or
-
sel
ect
iv
e
com
pr
essio
n
:
T
h
is
ki
nd
of
com
pr
essio
n
is
achiev
ed
thr
ough
sp
eci
fic
se
ct
or
of
im
age
(
I
)
and
is
nee
d
to
go
l
os
sle
ss
c
om
pr
ession
wh
i
le
sect
or
s
ubj
e
ct
ed
to
lo
ssy
c
om
pr
essio
n.
T
his
com
bin
at
io
n
of
com
pr
essio
n
te
chn
i
qu
e
help
s to
exteac
t t
he
si
gn
i
ficances
of
bo
t
h
te
ch
nique
s w
hic
h
el
im
inate
s the issu
es
of
the
com
pr
essi
on
process
.
T
he
m
anu
al
sect
ion
of
the
s
ect
or
is
crit
ic
al
con
c
ern
an
d
it
ca
n
be
el
im
inate
d
by
form
ulati
ng
the
functi
on
f
2
(
x)
that
do
e
s
the
autom
at
ed
sect
or
sel
ect
ion
(
Step
-
3).
T
he
al
gorithm
app
li
es
involu
ntary
th
r
esh
old
in
g
on
i
m
age
(
I
)
an
d
t
hen
c
on
ver
ts
it
in
to
bin
ary
for
m
.
The
proces
s
res
ults
an
im
age
(
I
4
)
after
m
ultip
li
cat
ion
of th
r
esh
old
im
age (The
process
of
this p
ro
ce
ss is i
ll
us
trat
ed
i
n
Fi
gure
4)
.
G
r
a
y
s
c
a
l
e
I
m
a
g
e
G
l
o
b
a
l
i
m
a
g
e
t
h
r
e
s
h
o
l
d
i
n
g
M
u
l
t
i
p
l
y
O
p
e
r
a
t
i
o
n
O
b
t
a
i
n
S
e
l
e
c
t
i
v
e
S
e
c
t
o
r
Figure
4
.
Stage
s of
s
ect
or
s
el
ect
ive
c
om
pr
ess
ion
d.
Segm
entat
ion
-
o
pe
rati
on:
To
the
sel
ect
ed
i
m
age
sect
ors
(
I
4
)
,
the
segm
entat
ion
process
is
app
li
ed
(S
te
p
-
4)
wh
e
re
ex
plici
t
functi
on
f
3
(
x)
is
us
ed
f
or
m
ulti
-
scal
e
seg
m
entat
ion
ope
rati
on.
The
functi
on
“
f
3
(
x)
”
us
es
fu
zzy
c
-
cl
ust
erin
g
f
or
bi
nar
iz
ed
im
age
“
I
4
”
.
Lat
er,
s
umm
a
tio
n
of
al
l
the
pi
xel
la
bels
is pe
rfor
m
ed.
Fi
nally
,
the
segem
ented
i
m
age
par
ts
are
com
bin
ed
thr
ough
m
ulti
plica
ti
on
and
it
will
helps
to
gen
e
rate
the
segm
ented
im
a
ge
(
BW
)
(
The
segem
entat
ion
process
is re
presented
in Fi
gu
re
5)
.
S
e
l
e
c
t
i
v
e
S
e
c
t
o
r
L
a
y
e
r
-
S
e
g
m
e
n
t
a
t
i
o
n
S
e
g
m
e
n
t
e
d
I
m
a
g
e
Figure
5
.
Stage
s of
s
e
gm
entat
i
on ope
rati
on
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Novel
h
y
br
id
fr
am
ew
or
k f
or
i
mage
c
omp
res
sion f
or
sup
po
r
ti
ve ha
r
dwar
e
desig
n of
..
.
(
P
remach
and D
.
R
.
)
1989
e.
Con
t
our
d
et
ect
ion
:
D
ur
i
ng
t
his
sta
ge,
a
m
at
r
ix
is
form
ed
to
retai
n
the
co
r
ner
m
et
ric
ob
a
ined
f
or
the
in
pu
t
i
m
age
and
is
con
si
der
e
d
f
or
f
eat
ur
e
ext
racti
on
(S
te
p
-
5)
.
A
n
inde
x
is
co
nst
ru
ct
ed
a
fter
fi
nd
i
ng
t
he
posit
ive
corner
set
s.
F
urt
her,
the
in
puts
and
outp
uts
are
m
at
ched
to
f
or
m
the
red
(r
1
),
gr
ee
n
(
g
1
),
an
d
bl
ue
(b
1
)
com
po
ne
nts.
T
he
“
r1
”
com
ponen
ts
are
i
niti
at
ed
with
25
5
a
nd
ot
her
com
po
ne
nts
li
ke
“g
1”
a
nd
“b
1”
a
r
e
i
nital
iz
ed
with
0.
F
ur
t
her,
the
se
com
po
ne
nts
are
co
ncatenat
ed
to
get
the
a
ct
ual
corner
po
ints
“R
”.
D
ur
i
ng
each
sta
ge
m
a
xim
u
m
reg
io
ns
of
the
i
m
age
are
identifie
d
t
o
detect
the
pe
aks
of
the
corners.
Furthe
r,
the
input
an
d
outp
uts
are
m
at
ched
by
w
hich
t
he
ne
w
c
om
po
ne
nts
of
t
he
“r”
,
“g”
an
d
“
b”
com
po
ne
nts
ar
e
con
cat
e
nated
to
get
the
sup
presse
d
c
orner
(
C
r
)
points
(
Ste
p
-
6).
T
he
c
ontour
detect
ion
al
gorithm
is
als
o
us
e
d
to
m
ulti
pl
y
the
gr
ay
scal
e
i
m
age
with
th
e
con
t
our
in
de
x
so
t
hat
the
c
on
t
ours
of
c
omplet
e
i
m
ahe
ca
be
gen
e
rated
(
Sta
ges of t
his im
ple
m
entat
ion
is
giv
e
n
in
Fig
ure
6)
.
S
e
g
m
e
n
t
e
d
I
m
a
g
e
C
o
n
t
o
u
r
M
a
t
r
i
x
F
i
n
d
R
e
d
G
r
e
e
n
B
l
u
e
C
o
m
p
o
n
e
n
t
s
C
o
n
d
i
t
i
o
n
f
o
r
p
o
s
i
t
i
v
e
c
o
n
t
o
u
r
s
c
o
n
c
a
t
e
n
a
t
i
o
n
O
b
t
a
i
n
c
o
n
t
o
u
r
p
o
i
n
t
s
E
x
t
r
a
c
t
M
a
x
i
m
i
z
e
d
r
e
g
i
o
n
I
n
d
e
x
c
o
u
n
t
o
u
r
s
m
u
l
t
i
p
l
i
c
a
t
i
o
n
S
u
p
p
r
e
s
s
e
d
C
o
n
t
o
u
r
P
o
i
n
t
s
Figure
6
.
Stage
s of
c
on
t
our
d
e
te
ct
ion
f.
Op
ti
m
iz
ed
cont
ours
detect
io
n
:
The
pe
rfo
rm
ance
of
the
co
ntour
detect
io
n
is
opti
m
ed
i
n
this
par
t
of
the
i
m
ple
m
entat
io
n.
D
ur
i
ng
this,
the
pr
i
or
inf
orm
ation
of
co
ntour
m
at
rix
will
be
acce
s
s
ed
a
nd
the
n
a
n
S
-
sh
a
ped
c
urve
functi
on
is
pr
oducin
g
the
i
nd
i
vidual
m
atr
ix.
T
he
valu
es
betwee
n
prob
a
bili
ty
lim
it
is
colle
ct
ed
for
the
co
ncatenati
on
of
fi
nal
co
nt
ours
le
adin
g
to
optim
iz
ed
co
ntour
po
i
nts
(
Cp)
(
Step
-
7)
.
The
final
pr
ocess
of
com
pr
essi
on
is
to
ap
ply
wa
velet
s
w
her
e
f
un
ct
io
n
f
4
(
x)
is
co
ns
ide
red
ov
er
the
im
age
w
it
h
con
t
our
po
i
nts
(Cp)
.
T
he
ope
r
at
ion
hel
ps
to
gen
e
rate
the
final
f
or
m
of
co
m
pr
essed
im
ag
e
ϕ
(
Step
-
8)
a
nd
this p
ro
ce
ss is
sh
ow
n
in
Fi
gur
e 7
.
S
e
g
m
e
n
t
e
d
I
m
a
g
e
S
-
s
h
a
p
e
d
c
u
r
v
e
m
e
m
b
e
r
s
h
i
p
f
u
n
c
t
i
o
n
A
p
p
l
y
c
o
m
p
u
t
a
t
i
o
n
a
l
g
e
o
m
e
t
r
y
P
e
r
f
o
r
m
w
a
v
e
l
e
t
c
o
m
p
r
e
s
s
i
o
n
Figure
7
.
Stage
s of
o
ptim
iz
ed
c
on
t
ours
d
et
ect
ion
3.
RESU
LT
A
N
ALYSIS
The
c
or
e
idea
of
th
e
outc
om
e
analy
sis
is
to
ens
ur
e
that
pro
po
s
ed
c
om
pu
ta
ti
on
al
m
od
el
sh
ould
offe
r
stream
li
ned
fa
ci
li
ta
t
ion
f
or
t
he
ha
r
dw
a
re
r
eal
iz
at
ion
of
the
im
age
com
pr
essi
on.
A
pa
rt
of
the
im
a
ge
i
s
consi
der
e
d
f
or
the
lossless
com
pr
essio
n
ov
er
dif
fer
e
nt
form
s
of
i
m
ages.
The
vis
ual
r
esults
of
the
f
il
te
red
i
m
age
(af
te
r
a
pp
ly
in
g
no
n
-
li
near
filt
er)
,
se
ct
or
im
age,
m
ulti
-
scal
e
seg
m
ented
i
m
age,
con
t
our
de
te
ct
ion
an
d
op
ti
m
iz
ed
counter
is
gi
ven
i
n
F
igure 8
.
T
he
analy
sis
of
t
he
syst
e
m
is
per
form
ed
fo
r
both selec
ti
v
e
reg
i
ons
a
nd
for
com
plete
i
m
age
by
con
s
iderin
g
Me
an
Square
Er
r
or
(
MSE),
Pea
k
S
ign
al
-
to
-
No
ise
Ra
ti
o
(P
SN
R
)
and
Com
pr
ession
Ra
ti
o
(as rep
re
sented
in
F
ig
ure
s
9
-
11
res
pecti
vely
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021
:
19
85
-
1993
1990
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
Figure
8
.
Vis
ua
l
o
utc
om
es o
f
s
am
ple
i
m
age
, (a)
o
rigin
al
i
m
age
, (
b)
filt
ere
d
i
m
age
, (
c)
se
le
ct
ed
s
ect
or
,
(d)
se
gm
ented
i
m
age
, (
e)
c
on
t
our
p
oin
ts
, (f
)
f
uzzy co
r
ner
po
ints
, (g
)
op
ti
m
i
zed
c
om
pr
ession
,
(h)
wav
el
et
c
om
pr
ession
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Novel
h
y
br
id
fr
am
ew
or
k f
or
i
mage
c
omp
res
sion f
or
sup
po
r
ti
ve ha
r
dwar
e
desig
n of
..
.
(
P
remach
and D
.
R
.
)
1991
Figure
9
.
PS
N
R
p
er
f
or
m
ance
Figure
10
.
Me
a
n
s
quare
d
e
rror
p
e
rfor
m
ance
Figure
11
.
C
om
pr
ession
r
at
io
The
a
naly
sis of ab
ov
e
facto
rs
i
nd
ic
at
es t
hat th
e PSN
R
pe
rform
ance o
f
syst
em
is im
pr
ov
e
d i
n
SR th
a
n
FI
.
T
he
sign
al
qu
al
it
y
of
the
syst
e
m
al
so
i
m
pr
ov
e
d
at
reduce
d
dim
e
ns
io
n.
T
he
al
gorithm
ta
kes
about
0.987
7s
ecs
t
o
respo
nd
to
t
he
exec
ution
w
hi
ch
is
c
onside
re
d
as
i
ns
ta
nta
ne
ou
s
to
offe
r
c
ost
eff
ect
ive
so
l
ution
for
the
co
m
pr
e
ssion o
pe
rati
ons.
4.
CONCL
US
I
O
N
This
pa
per
ha
s
pr
ese
nted
a
discuss
i
on
of
the
com
pu
t
at
ion
al
m
od
el
fo
r
facil
it
at
i
ng
hard
war
e
acce
le
rati
on
w
hile
pe
rfor
m
ing
im
age
com
pr
ession.
T
he
si
gnific
ant
c
ontri
bu
ti
on
of
this
pap
e
r
is
t
o
br
i
dg
e
the
trade
-
off
be
we
en
the
ha
r
dw
a
re
-
base
d
a
nd
s
of
t
war
e
-
base
d
com
pr
essio
n
a
ccel
erati
on
pro
cess
with
res
pe
ct
to
the
VLSI
archi
te
ct
ur
e.
The
c
on
t
rib
ution
s
of
the
pr
op
os
ed
syst
e
m
towards
the
i
m
age
com
pr
essio
n
are
as
fo
ll
ows:
i)
the
pro
po
se
d
syst
em
is
cost
ef
fec
ti
ve
as
it
does
n’t
us
e
com
plete
res
ource
f
or
com
pr
essio
n
a
s
only
the
sel
ect
ive
se
ct
or
is
s
ubj
e
ct
ed
f
or
the
lossle
ss
com
pr
essio
n,
ii
)
the
pro
pos
ed
syst
em
is
a
l
so
c
os
t
ef
fecti
ve
as
the
c
om
pr
essio
n
is
ca
rr
ie
d
ou
t
over
t
he
li
ght
weig
ht
c
on
t
our
featu
re
t
hat
not
only
offer
s
stora
ge
optim
i
zat
ion
bu
t
al
s
o
retai
ns
bette
r
si
gn
al
qu
al
it
y,
an
d
i
ii
)
the
im
ple
mentat
ion
of
the
pro
po
se
d
syst
e
m
do
es
n’
t
in
cl
ude
m
axi
m
u
m
d
ep
end
e
nc
ie
s
of r
e
so
urce t
o
ca
rr
y
out the tas
k.
REFERE
NCE
S
[1]
T.
Lin
and
P
.
Hao
,
“
Com
poun
d
image
compress
ion
for
rea
l
-
t
ime
comp
ute
r
scre
en
image
tr
a
nsm
ission
,”
IEE
E
transacti
ons on
I
mage
Proce
ss
in
g
,
vol
.
14
,
no
.
8
,
pp.
993
–
1005
,
2
005.
[2]
H.
G.
M
y
ung,
J
.
Li
m
,
and
D
.
J.
Goodm
an
,
“
Single
ca
rri
er
F
DM
A
f
or
upli
n
k
wire
l
ess
tra
ns
m
ission
,”
IEEE
Ve
hi
cul
ar Tec
hn
ology
Magazin
e
,
vol.
1,
no.
3
,
pp.
30
–
38
,
2006.
[3]
N.
Java
id,
Z
.
Abbas,
M.
S.
Fare
ed,
Z
ahoor
Ali
Khan,
and
N.
Alra
je
h
,
“
M
-
ATTE
MP
T:
A
new
ene
rg
y
-
eff
ic
i
en
t
routi
ng
pro
toc
ol
for
wire
l
ess bod
y
area
sensor
netw
orks
,”
Proc
edia Compute
r Sc
ience
,
vol
.
19
,
pp
.
224
–
231
,
2013
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
11
, No
.
3
,
J
une
2021
:
19
85
-
1993
1992
[4]
B.
Guo,
D
.
Zha
ng,
Z
.
W
ang,
Z
.
Yu,
and
X
.
Zhou
,
“
Opportunisti
c
IoT:
Expl
or
ing
the
har
m
onious
int
era
ction
bet
wee
n
hu
m
an
and
th
e
int
ern
et
of
thi
ngs
,”
J
ournal
of
Ne
tw
ork
and
Computer
Applica
ti
ons
,
vol
.
36
,
no.
6
,
pp.
1531
–
1539
,
2013.
[5]
L.
At
zor
i
,
A
.
Ier
a,
and
G
.
Morabi
to
,
“
Siot:
Giving a s
oci
al
struc
tur
e to the int
ern
et
of things
,”
IEE
E
communic
ati
on
s
le
tters
,
vol
.
15
,
n
o.
11
,
pp
.
1193
–
1195
,
2011
.
[6]
R.
M.
Dy
as
and
B
.
Ta
ng
,
“
Software
,
m
et
hod
a
nd
appa
rat
us
for
rat
e
cont
rol
le
d
image
compress
ion
,”
U.S.
Pate
n
t
6,
504,
494
,
Jan
.
7,
2003
.
[7]
S.
J.
W
ee
and
M
.
P.
Schu
y
l
er
,
“
Im
age
compre
ss
ion
fea
tur
ing
sele
c
ti
ve
re
-
use
of
prior
compress
ion
dat
a
,”
U
.
S.
Pate
nt
6,
697
,
061
,
Feb
.
24
,
2004
.
[8]
S.
Chatterjee
an
d
S
.
Sen
,
“
Cac
h
e
-
eff
i
cient
m
at
r
i
x
tra
nspos
ition
,”
i
n
Proceedi
ngs
Six
th
Inte
rnat
ion
al
Symposium
o
n
High
-
Pe
rform
ance
Comput
er
Ar
chi
t
ec
ture
,
HP
C
A
-
6
(Cat
.
No.
PR
00550),
2000
,
pp.
195
–
205
.
[9]
J.
L.
Ben
tl
e
y
,
D.
D.
Sl
ea
tor
,
R
.
E
.
Ta
rj
an,
a
nd
V
.
K.
W
ei
,
“
A
loc
a
lly
ad
a
pti
ve
data
compress
ion
sche
m
e
,”
Comm
unic
ati
ons of
the A
CM
,
vol
.
29
,
no
.
4
,
pp
.
3
20
–
330
,
1986
.
[10]
A.
Jas,
J
.
Ghos
h
-
Dastida
r, M
.
E
.
Ng,
and
N
.
A.
T
ouba
,
“
An e
ffi
cient
t
est
ve
ct
or
co
m
pre
ss
ion
sche
m
e
using se
le
c
tive
Huffm
an
codi
ng
,”
IE
EE
tr
ansact
ions
on
computer
-
aide
d
design
o
f
int
egrat
ed
ci
rc
uit
s
and
systems
,
vol.
22
,
no.
6
,
pp.
797
–
806
,
20
03
.
[11]
M.
Herná
nde
z
-
C
abr
oner
o,
V.
Sa
nche
z
,
I.
B
la
n
es,
F.
Aulí
-
Llinàs,
M.
W
.
Marc
elli
n
and
J.
Serra
-
S
agr
istà,
“
Mos
ai
c
-
Based
Color
-
Tr
ansform
Optimi
za
t
ion
for
Lossy
and
Loss
y
-
to
-
Lossless
Com
pr
ession
of
Patholog
y
W
hole
-
Slid
e
Im
age
s,
”
in
I
EEE
Tr
ansacti
ons
on
Me
di
cal Imaging
,
vo
l. 38, no. 1, pp. 21
–
32,
Ja
n.
2019
.
[12]
K.
Dim
il
il
er
and
A.
A
m
irc
anov,
“
Im
age
Com
pre
ss
ion
Sy
stem
with
an
Op
ti
m
isat
i
on
of
Com
pre
s
s
ion
Rat
io
,”
IET
Image
Proc
essing
,
vol
.
13
,
2019
.
[13]
H.
W
ang,
T
.
W
ang,
L.
Li
u
,
H.
Sun
,
and
N.
Zhe
ng,
“
Eff
ic
i
ent
Com
pre
ss
ion
-
Based
Li
n
e
Buffe
r
Design
for
Im
age
/Vide
o
Pr
oce
ss
ing
Circ
u
its
,
”
in
IEEE
Tr
ansacti
ons
on
V
ery
Lar
ge
Scale
In
te
gration
(
VLSI)
Syste
ms
,
vo
l.
2
7
,
no.
10
,
pp
.
2423
–
2433,
Oct
.
201
9.
[14]
C.
Chen,
S.
Ch
en,
C.
Li
o
a
,
an
d
P.
A.
R.
Abu
,
“
Lossless
CF
A
Im
age
Com
p
ression
Chip
De
sign
for
W
ire
le
s
s
Capsule
Endosc
op
y
,
”
in
I
EE
E
A
cc
ess
,
vol. 7, pp. 107047
–
107057
,
2019
.
[15]
Y.
Hala
wan
i,
B.
Moham
m
ad,
M.
Al
-
Quta
y
r
i
,
an
d
S.
F.
Al
-
Sar
a
wi,
“
Mem
ristor
-
Based
Hardware
Acc
e
le
ra
tor
for
Im
age
Com
pre
s
sion,
”
in
IE
EE
Tr
ansacti
ons
on
Ve
ry
Lar
ge
Scal
e
Inte
grat
ion
(
VLSI
)
Syste
ms
,
vol.
26,
no.
12
,
pp.
2749
–
2758
,
Dec
.
2018.
[16]
J.
Choi,
B.
Kim
,
H.
Kim
,
and
H.
Le
e,
“
A
High
-
Throughput
Har
dware
Acc
e
le
r
ator
for
Lossless
Com
pr
ession
of
a
DD
R4
Comm
an
d
Tra
c
e,
”
in
I
E
EE
Tr
ansacti
ons
on
Ve
ry
Lar
ge
Scal
e
In
te
gratio
n
(
VLSI
)
Syste
ms
,
vol.
27
,
no.
1
,
pp.
92
–
102
,
Jan
.
2019
.
[17]
A.
Kaur,
D.
Mishra,
S.
J
ai
n
,
and
M.
Sark
ar,
“
Cont
ent
D
rive
n
On
-
Chip
Com
pre
ss
ion
and
Ti
m
e
Eff
i
ci
e
nt
Rec
onstruc
ti
on
f
or
Im
age
Sensor
Applicati
ons,
”
i
n
IEEE
S
ensors
Journal
,
vol
.
1
8,
no
.
22
,
pp.
9169
–
9179
,
Nov.
2018.
[18]
T.
Onishi
et
al
.
,
“
A
Single
-
Chip
4K
60
-
fps
4
:2:
2
HEVC
Video
Enc
od
er
L
SI
Emplo
y
ing
Eff
icient
Motio
n
Esti
m
at
ion
and
Mode
Dec
ision
Fram
ework
W
it
h
Scal
ability
to
8K,
”
in
IE
EE
Tr
ansacti
ons
on
Ve
ry
Lar
ge
Scale
Inte
gration
(
VLSI)
Syste
ms
,
vol
.
26,
no
.
10
,
pp
.
1
930
–
1938,
Oct
.
2018.
[19]
Z.
H.
Z
hu,
X.
Meng,
J.
Zhou
,
and
B.
Z
eng,
“
C
om
pre
ss
ion
-
Depe
ndent
Tra
nsfor
m
-
Do
m
ai
n
Dow
nward
Conversi
o
n
for
Bloc
k
-
Based
I
m
age
Coding,
”
in
IEEE
Tr
ansacti
ons
on
Image
Proce
ss
ing
,
vol
.
27,
no
.
6,
pp
.
2635
–
2649,
Jun
.
2018
.
[20]
J
.
Hs
ie
h,
M.
Shih
,
and
X.
Huang,
“
Algorit
hm
and
VLSI
Archi
te
ct
ur
e
Design
of
Low
-
Pow
er
S
P
IHT
Dec
oder
for
m
Hea
lt
h
Appli
c
at
ions,
”
in
I
EEE
Tr
ansacti
ons
on
Bi
omedi
cal
C
i
rcuit
s
and
Syst
e
ms
,
vol.
12
,
no.
6,
pp.
1
450
–
145
7
,
Dec
.
2018
.
[21]
Q.
Chen,
H.
Su
n
,
and
N.
Zhe
ng
,
“
W
orst
Case
D
rive
n
Displ
a
y
Fr
ame
Com
pre
ss
ion
for
En
erg
y
-
Eff
ic
i
e
nt
Ul
tra
-
HD
Displa
y
Proce
ss
i
ng,
”
in
I
EE
E
Tr
ansacti
ons on
M
ult
imedi
a
,
vo
l. 2
0,
no
.
5
,
pp
.
111
3
–
1125,
Ma
y
.
2
018.
[22]
K.
Kim
,
C.
Le
e
,
and
H.
Le
e
,
“
A
Sub
-
Pixel
Gradi
ent
Com
pre
ss
ion
Algorit
hm
for
Te
xt
Im
age
Dis
play
on
a
Sm
art
Devic
e
,
”
in
IEEE
Tr
ansacti
ons
on
Consum
er
Elec
troni
cs
,
vo
l. 6
4,
no
.
2
,
pp
.
231
–
239,
Ma
y
.
201
8.
[23]
L.
F.
R.
Lu
ca
s, N
.
M.
M.
Rodrigue
s,
L.
A.
da
Sil
va
Cruz
,
and
S. M
.
M.
de
Faria
,
“
Lossless Co
m
p
ression
of
Medical
Im
age
s
Us
ing
3
-
D
Predic
tors,
”
i
n
IEEE
Tr
ansacti
ons
on
M
edi
ca
l
Imaging
,
vol
.
36,
no.
11,
pp.
2250
–
2260,
Nov
.
2017.
[24]
S.
Yin,
P.
Ou
y
a
ng,
T.
Ch
en,
L
.
Li
u
,
and
S.
W
ei
,
“
A
Configur
abl
e
Par
al
l
el
Ha
rdware
Archi
t
ecture
for
Eff
ic
i
en
t
Inte
gra
l
Histogr
am
Im
age
Com
puti
ng,
”
in
IE
E
E
Tr
ansacti
ons
on
Ve
ry
Lar
ge
Scal
e
Inte
grat
io
n
(
VLSI)
Syste
ms
,
vol.
24
,
no
.
4
,
pp
.
1305
–
1318
,
Ap
r
.
2016
.
[25]
S.
S.
Parikh
,
D.
Ruiz,
H.
K
al
va
,
G.
Fern
án
dez
-
Escr
iba
no
,
and
V.
A
dz
ic
,
“
High
Bit
-
Dept
h
Medic
a
l
Im
a
ge
Com
pre
ss
ion
Wi
th
HEVC,
”
in
I
EE
E
Journal
of
Bi
omedi
cal
and
Healt
h
Informat
ic
s
,
vol
.
22
,
no.
2,
pp.
552
–
560,
Mar
.
2018
.
[26]
S.
Chen,
T.
Li
u
,
C.
She
n
,
and
M.
Tua
n
,
“
VLSI
Im
ple
m
ent
atio
n
of
a
Cost
-
Eff
i
ci
en
t
Nea
r
-
Lossless
CF
A
Im
ag
e
Com
pre
ss
or
for
W
ire
le
ss
Capsul
e
Endoscop
y
,
”
i
n
IEEE Access
,
vol.
4
,
pp
.
10235
–
10245,
2016
.
[27]
P.
Pete
r,
L
.
Kaufhol
d
,
and
J.
W
ei
cke
rt
,
“
Turni
ng
Diffusion
-
Based
Im
age
Colori
z
at
ion
Int
o
Eff
ic
i
ent
Col
or
Com
pre
ss
ion,
”
i
n
IEEE
Tr
ansact
ions o
n
Image
P
roce
ss
ing
,
vol
.
2
6,
no
.
2
,
pp
.
860
–
869,
Feb
.
2017
.
[28]
S.
Zha
ng,
X.
Ti
a
n,
C.
Xiong
,
and
J.
Ti
an,
“
Unified
VLSI
arc
hit
ect
ure
for
photo
cor
e
tra
nsform
used
in
JP
EG
XR,
”
in
E
le
c
tronic
s L
et
t
ers
,
vol
.
51
,
n
o.
8
,
pp
.
628
–
63
0,
Apr.
2015.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Novel
h
y
br
id
fr
am
ew
or
k f
or
i
mage
c
omp
res
sion f
or
sup
po
r
ti
ve ha
r
dwar
e
desig
n of
..
.
(
P
remach
and D
.
R
.
)
1993
[29]
S.
M.
A.
Zeinolabedi
n,
J.
Zhou
,
X.
Li
u
,
and
T
.
T
.
Kim
,
“
An
Area
-
and
Ene
rg
y
-
E
f
fic
i
ent
FIF
O
De
sign
Us
ing
Err
or
-
Reduc
ed
Da
ta
C
om
pre
ss
ion
and
Nea
r
-
Thre
shold
Opera
ti
o
n
for
I
m
age
/Vide
o
Ap
pli
c
at
ions,
”
in
I
EE
E
Tr
ansacti
o
ns
on
Ve
ry
Lar
ge
S
cal
e
Inte
grat
ion (
VLSI
)
Syste
ms
,
vol.
23
,
no
.
11
,
p
p.
2408
–
2416
,
N
ov.
2015
.
[30]
S.
Kim
,
M.
Ki
m
,
J.
Kim
,
and
H.
L
ee,
“
Fixe
d
-
Rat
io
Com
pre
ss
ion
of
an
R
GBW
Im
age
a
nd
Its
Hardwar
e
Im
ple
m
ent
at
ion
,
”
in
IEEE
Journal
on
Em
ergin
g
and
Se
le
c
te
d
Topics
in
Circu
it
s
and
S
yste
ms
,
vol
.
6
,
no
.
4,
pp.
484
–
496
,
De
c.
2016
.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Pr
em
a
chand
D
.
R
.
,
cur
r
ent
l
y
working
as
As
s
oci
a
te
Profess
or,
Depa
rtment
of
El
ectroni
cs
and
Com
m
unic
at
ion
Engi
ne
eri
ng
,
Ba
ll
ari
Instit
u
te
of
Te
chno
log
y
and
Mana
gement,
Karna
t
aka,
Indi
a
.
He
has
industr
i
al
expe
ri
ence
o
f
2
y
e
ars
and
t
ea
ch
ing
exp
e
ri
e
nce
of
17
y
e
ars
.
He
has
ne
arly
publi
shed
3
jou
r
nal
s
and
3
conf
e
ren
ce
pape
r
.
His
areas
of
intere
st
are
in
VLSI,
Im
age
pro
ce
ss
ing
and
Micro
e
le
c
tr
onic
s.
Dr.
U.
Erann
a
,
cur
ren
t
l
y
worki
ng
as
Profess
or
and
HO
D
in
the
Depa
rtment
of
El
e
ct
roni
cs
and
Com
m
unic
at
ion
Engi
ne
eri
ng,
Ba
l
la
ri
Instit
u
te
of
Te
chno
log
y
and
Mana
gement.
H
e
has
publi
shed
8
conf
ere
n
ce
pa
per
s
and
20
journa
l
papers
.
His
are
a
of
rese
arch
was
Comm
un
ic
a
ti
on,
Con
t
ro
l
S
y
stem.
Evaluation Warning : The document was created with Spire.PDF for Python.