An improved human visual system based reversible data hiding ...

yakzephyrΤεχνίτη Νοημοσύνη και Ρομποτική

24 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

58 εμφανίσεις

1

An improved human visual system based
reversible data hiding
method
using adaptive
histogram modification

Source: Optics Communications
, VOL. 291,
pp

87
-
97,
March
2013


Authors:
Wien
Hong, Tung
-
Shou

Chen, Mei
-
Chen
Wu

Speaker
: Chi
-
Hsien

Lin


Date:2013/04/03

1

Outline

Introduction

Proposed Scheme

Experimental Result

Conclusions

2

Introduction (1/4)

MED







3

Introduction (2/4)

edge
pixels (1) and
non
-
edge
pixels (0)







4

Introduction (3/4)

HVS:

H
uman
V
isual
S
ystem









[13]

Weisi

Lin
,

Li Dong, and Ping
Xue
,

Visual
Distortion Gauge Based on Discrimination of

Noticeable Contrast
Changes”,
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO
TECHNOLOGY, VOL. 15, NO. 7, JULY
2005

5

a
=
10
,
b=
20
, and
c=
24
when
E
i
,
j
=
0

(
for
non
-
edge
pixels)



a
=
8
,
b=
18
, and
c=
22
when
E
i
,
j
=
1

(
for edge
pixels)


Introduction (4/4)

JND


Just Noticeable Difference

A
bigger JND signals more distortion tolerance

of the
HVS




where
λ
=0.5 ,
TA
i,j
(x) is the difference value
between the largest and the smallest pixel
values
in

i,j



[28] N.
Jayant
, J. Johnston, and R.
Safranek
, “Signal compression based
on models
of human
perception,”
Proc. IEEE
, vol. 81, pp. 1385

1422,
Oct.1993
.




6

Proposed Scheme (1/10)
-

Embed









7

Input: Cover image î,
maximum embedding level
L
Max
, and secret data S.

Output:
Stego

image I’,
the length of compressed location map C
LM
,
and
the length of secret data S.


Step1:
Pre
-
shift the cover image to obtain a pre
-
shifted
image
I
and
a compressed location map
C
LM


(1)
Calculate the embedding level
L
i,j

for
I
i,j








Proposed Scheme
(2/10)
-

Embed

155

155

245

238

151

151

160

154

251

241

151

154

155

155

155

156

8

Ex:
L
max
=5 , TE=100, I
2,2
=151,
i
=2,j=2

Average(Ω)=(155+155+151)/3=153.7

Var
(Ω)=(155
2
+155
2
+151
2
-
3*153.7
2
)/3=3.56 < TE

Ω is smooth region


(JND)
J
i,j
=
TL
i,j
(x)+0.5TA
i,j
(x)/
TL
i,j
(x)



=12.8+0.5*4/12.8=12.96


Where
TL
i,j
(x)

=(24
-
10)/(255
-
125)*(151
-
125
)+10=12.8



k=0 , 2
0
<
J
i,j

, 2
5
<Average(Ω)<255
-
2
5

, 0 ≤
L
max


k=1
,
2
1
<
J
i,j

,
2
5
<Average(Ω
)<
255
-
2
5

,
1

L
max


k=2
,
2
2
<
J
i,j

,
2
5
<Average(Ω
)<
255
-
2
5

,
2

L
max


k=3
,
2
3
<
J
i,j

,
2
5
<Average(Ω
)<
255
-
2
5

,
3

L
max


k=4
,
2
4
>
J
i,j

,
2
5
<Average(Ω
)<
255
-
2
5

,
4

L
max


L
i,j
=3 and 2
3


I
2,2


255
-
2
3


LM
i,j
=0


Proposed Scheme
(3/10)
-

Embed

9

Ex:
L
max
=5 , TE=100, I
2,3
=160,
i
=2,j=3

Average(Ω)=(155+155+151)/3=183.7

Var
(Ω)=(155
2
+245
2
+151
2
-
3*183.7
2
)/3=1883 > TE

Ω is complex region


(JND)
J
i,j
=
TL
i,j
(x)+0.5TA
i,j
(x)/
TL
i,j
(x)



=11.77+0.5*94/11.77=15.76


Where
TL
i,j
(x)

=(
22
-
8)/(
255
-
125
)*(160
-
125)+8=11.77



k=5 , 2
5
>
J
i,j

, 2
5+1
<Average(Ω)<255
-
2
5+1

, 5 ≤
L
max


k=4
,
2
4
>
J
i,j

,
2
5+1
<Average(Ω
)<
255
-
2
5+1

,
4≤
L
max


k=3
,
2
3
<
J
i,j

,
2
5+1
<Average(Ω
)<
255
-
2
5+1

,
3≤
L
max




L
i,j
=4 and 2
4


I
2,3


255
-
2
4


LM
i,j
=0


155

155

245

238

151

151

160

154

251

241

151

154

155

155

155

156

Proposed Scheme
(4/10)
-

Embed

1
0

Ex:
L
max
=5 , TE=100, I
2,3
=240,
i
=2,j=3

Average(Ω)=(155+155+151)/3=183.7

Var
(Ω)=(155
2
+245
2
+151
2
-
3*183.7
2
)/3=1883 > TE

Ω is complex region


(JND)
J
i,j
=
TL
i,j
(x)+0.5TA
i,j
(x)/
TL
i,j
(x)



=20.38+0.5*94/20.38=22.68


Where
TL
i,j
(x)

=(
22
-
8)/(
255
-
125
)*(240
-
125)+8=20.38



k=5 , 2
5
>
J
i,j

, 2
5+1
<Average(Ω)<255
-
2
5+1

, 5 ≤
L
max


k=4
,
2
4
<
J
i,j

,
2
5+1
<Average(Ω
)<
255
-
2
5+1

,
4≤
L
max


k=3
,
2
3
<
J
i,j

,
2
5+1
<Average(Ω
)<
255
-
2
5+1

,
3≤
L
max




L
i,j
=5 and I
2,3

≥ 255
-
2
4


LM
i,j
=1 and
I
I,j
=240
-
2
5
=208


155

155

245

238

151

151

240

154

251

241

151

154

155

155

155

156

155

155

245

238

151

151

208

154

251

241

151

154

155

155

155

156

Proposed Scheme (5/10)
-

Embed









1
1

Compress LM
using JBIG2 coder to get a
compressed
location map C
LM
.


Concatenate C
LM

and
secret data
S to get
M=C
LM
|S

Proposed Scheme (6/10)
-

Embed









1
2

Use MED to get the prediction
error
d
i,j

=
I
i,j
-
p
.

If
-
2
Li,j
≤d
i,j
<2
Li,j
,
extract a bit
m from M,
and
use the
following
equation to embed
m into
I’
I,j
=
I
I,J
+d
i,j
+m


Otherwise
,
I
i,j

is
shifted by using the following
equation

Proposed Scheme (7/10)


Embed









1
3

Let M=1011…

i
=2,j=2

p=151,d
i,j

=
I
i,j
-
p=151
-
151=0.
L
i,j
=3

If
-
2
Li,j
≤d
i,j
<2
Li,j
,

I’
I,j
=
I
I,J
+d
i,j
+m
=151+0+1=152


i
=2,j=3

p=242,d
i,j

=
I
i,j
-
p=160
-
242=
-
82.
L
i,j
=4

-
2
Li,j
≤d
i,j
<2
Li,j
, (
ow
)

I’
I,j
=I
I,J
-
2
4
=160
-
16=144



155

155

245

238

151

151

160

154

251

241

151

154

155

155

155

156

155

155

245

238

151

152

160

154

251

241

151

154

155

155

155

156

155

155

245

238

151

152

144

154

251

241

151

154

155

155

155

156

Proposed Scheme (8/10)
-

Extract









1
4

Proposed Scheme
(9/10)
-

Extract

Embed

d
i,j

=
I
i,j
-
p

If
-
2
Li,j
≤d
i,j
<2
Li,j
,

I’
I,j
=
I
I,J
+d
i,j
+m
= 2I
I,j
-
p+m


Extract

d’
i,j

=
I’
i,j
-
p=2I
I,j
-
p+m
-
p=2(
I
i,j
-
p)+m=2d
i,j
+m

mod(d’
i,j
,2)=mod(m,2)=m

I
i,j
=
p+d
i,j
=
p+floor
(
d’
i,j
/2)=
p+d’
i,j
-
ceiling
(
d’
i,j
/2
)



=
I’
i,j
-
ceiling(
d’
i,j
/2)



1
5

Proposed Scheme (10/10)
-

Extract









1
6

i
=2,j=2

L
i,j
=3,p=151

d

i,j

=
I’
i,j
-
p=152
-
151=1.

If
-
2
Li,j+1
≤d’
i,j
<2
Li,j+1
,

m=mod(d’
i,j
,2)=1 ,


I
i,j
=
d
i,j
+p
=151+[
d’
i,j
/2]=151


i
=2,j=3

L
i,j
=4,p=242,d’
i,j

=
I’
i,j
-
p=144
-
242=
-
98.
-----
(
-
2
Li,j+1
)≤
d’
i,j
<2
Li,j+1
, (
ow
)

I
I,j
=I’
I,J
+2
4
=144+16=160



155

155

245

238

151

151

160

154

251

241

151

154

155

155

155

156

155

155

245

238

151

151

144

154

251

241

151

154

155

155

155

156

155

155

245

238

151

152

144

154

251

241

151

154

155

155

155

156

Experimental Result

1
7

Experimental Result

1
8

Experimental Result

1
9

Experimental Result

2
0

Experimental Result

2
1

Experimental Result

2
2

Experimental Result

2
3

Experimental Result

2
4

2
5

A
new reversible data hiding method based
on the human visual system.


An
adaptive
pre
-
processing
method to
adaptively pre
-
shift the pixel that
improve the
Jung
et
al.’s
work.


The proposed
method offers better
performance over other existing works in
terms of payload and image quality.

2
6

Ref.

[17]
Seung
-
Won Jung, Le
Thanh

Ha, and Sung
-
Jea

Ko
, “
A New
Histogram Modification Based Reversible Data Hiding Algorithm
Considering the Human Visual System
µ??,(((?6,*1$/?352&(66,1*?
LETTERS, VOL. 18, NO. 2, FEBRUARY 2011

[11]W
. L. Tai, C. M.
Yeh
, and C. C. Chang, “Reversible data hiding based
on histogram modification of pixel differences,” IEEE Transactions on
Circuits and Systems for Video Technology, vol. 19, no. 6, pp. 906
-
910, 2009.


[8]P
. Tsai, Y.C. Hu, H.L.
Yeh

,”
Reversible
image hiding scheme using
predictive coding and histogram
shifting”

Signal
Processing, 89 (6)
(2009), pp.
1129

1143

[1] W
.

Hong

and T.S. Chen, “A Local Variance
-
Controlled Reversible
Data Hiding Method Using Prediction and Histogram
-
Shifting,”

The
journal of systems and
software, vol.83
, no. 12, pp. 2653
-
2663,
2010.




2
7

Steganalysis

2
8