A Novel Approach Towards Separable Reversible

Data Hiding Technique

Vinit Agham

Tareek Pattewar

Department of Computer Engineering

R C Patel Institute of Technology,Shirpur.
Dist.Dhule, Maharashtra, India
vinitagham@gmail.com

Department of Information Technology

R C Patel Institute of Technology,Shirpur.
Dist.Dhule, Maharashtra, India
tareekpattewar@gmail.com

Abstract- Internet is
medium

now

days

the

but

most

message

popular

communication

communication

over

the

internet is facing some problem such as data security, copyright

control, data size capacity, authentication etc. There are so many
research

is

progressing

on

the

field

like

internet

security,

steganography, cryptography. When it is desired to send the

confidential/important/secure

data

over

an

insecure

and

application of reversible data hiding technique is in IPR

(Intellectual Property Rights) protection, authentication. In
some application scenarios it is essential to provide security,
authentication and privacy during communication or
transferring data. That's why to hide the data or to provide the
data security we need some new approach in communication.

bandwidth-constrained channel it is customary to encrypt as well

as

compress

the

cover

data

and

then

embed

paper introduces the new way of originating the existing concept

i.e. separable reversible data hiding. Actually the concept of
separable

reversible

data

hiding

technique

is

based

on

steganography and related with internet security. The chief

objectives of this literature is to work on the concept in which we
used text as a hidden data, no plain spatial domain is used,
attempt to increase the amount of data which is to be hide,
evaluating quality by different interpretations. The principal

SEPARABLE REVERSIBLE DATA HIDING TECHNIQUE

II.

the

confidential/important/secure data into that cover data. This

The term "reversible data hiding" means getting the exact

recovery of the data after performing the process like
encryption-decryption and data hiding. Now the question is;
what is mean by "separable reversible data hiding technique"?
The word separable means it separates two major activities in
the scheme. These two activities are getting the exact recovery
of the secure hidden data and exact recovery of cover data
which is used to hide as shown in Fig.l.

notion of separable reversible data hiding is consist of three key

procedures. First encrypt the cover media second hide the data
and third get the data as well as cover media as per provisions.

Keywords- Image encryption, image recovery, reversible data

.II!!!!!!!II
!I ..l

hiding, separable reversible data hiding, embedded data,data-hiding

key,qllality index,PSNR.

Original

I.

'm .g. -...

OVERVIEW

Now a day the data security and data integrity are the two
challenging areas for research. There are so many research is
progressing on the field like internet security, steganography,
cryptography. Sometimes we found certain distortion in images
used in military, medical science which is un-acceptable.
Hence for data hiding we have a technique using which we can
extract data correctly and after that original cover content can
be perfectly recovered. This technique is known as reversible
data hiding. This technique is also called as lossless, distortion
free, or invertible data hiding technique [1]-[2]. Reference [3]
presented a unique reversible (Iossless) data hiding technique
which enables the exact recovery of the original signal with the
extraction of the embedded information. And this exact
recovery with lossless data is nothing but the reversible data
hiding. Usually the well-known LSB (least significant bits)
method is used as the data embedding method. Reversible data
hiding is a technique that is mainly used for the authentication
of data like images, videos, electronic documents etc. The chief

978-1-4799-2900-9/14/$31.00 2014 IEEE

Encrypted image containing embedded

Rec.overed
image

..w;;;..
s-..

'+-

"IU

extracted
bits

Fig.

L The scheme of separable reversible data hiding [4].

The Author in [4] uses image as a cover medium. The

scheme is divided into three steps in the first step; a sender or
content owner encrypts the cover image using an encryption
key. Then, a data-hider compresses the least significant bits of

771

the encrypted image using a data-hiding key to create enough

space to accommodate some additional data (Secure data).Now
at receiver side it occurs three cases as shown in Fig. 2 case
one is if a receiver has the data-hiding key only he can take out
the secure data even he does not know the cover media i.e.
cover image content. In second case if the receiver has the
encryption key only, he can decrypt the received data to obtain
an image similar to the original one, but he cannot extract the
additional data. If the receiver has both the keys i.e. data-hiding
key and the encryption key he can extract the additional data
and recover the original image. Thus we are separating the two
actions i.e. exact recovery of the hidden data and recovery of
the cover media. That's why it is termed as "Separable
Reversible Data Hiding Technique" which is reversible as well
as separable in nature.
III.

MOTIVATION AND RELATED WORK

Most of the reversible data hiding techniques are not

separable, not based on encryption-decryption domain, not
based on RGB-LSB steganography. Here presents the small
survey that reveals what techniques have been used for
compression-decompression,
encryption-decryption,
data
embedding etc.
Reversible data hiding technique includes following
actions
like
compression-decompression,
encryption
decryption, data embedding-data extracting, creating space at
LSB of the pixels of the image, providing security,
authentication using automatic key generation and etc. The
traditional way of transmitting redundant data is to first
compress the data to reduce the redundancy and then to
encrypt the compressed data. At the receiver side the
decryption and decompression operations are orderly
performed to recover the original cover data. However in
some applications a sender needs to transmit some data to the
receiver and wishes to keep the information confidential to a

network operator who provides the channel resource for the

transmission, means the sender should encrypt the original
data i.e. image in this case and the network provider compress
the encrypted data without any knowledge of the
cryptographic key and the original data. At receiver side to
reconstruct the original data, a decoder integrating
decompression and decryption functions will be used.
There
are
several
techniques
for
compressing/decompressing encrypted data have been
developed. When it is desired to transmit redundant data over
an insecure and bandwidth-constrained channel, it is usual to
first compress the data and then encrypt it. In [5] these steps
are in reverse order of that is first encrypting and then
compressing without compromising either the compression
efficiency or the security. They have examined the possibility
of first encrypting a data stream and then compressing it, such
that the compressor does not have knowledge of the
encryption key.
Reference [6] suggested a lossless compression method for
encrypted gray image using resolution progressive
decomposition in which there is much better coding efficiency
and less computational complexity. In this paper [6] resolution
progressive compression scheme was implemented which
compresses an encrypted image progressively in resolution
such that the decoder can observe a low-resolution version of
the image, studies the statistics based on it and use this
statistics to decode the next resolution level and this process
iterates until last level of resolution. Here the image
undergoes
stream-cipher
based
encryption
before
compression.
In paper [7] there is a buyer-seller watermarking protocol
which is the concept of digital watermarking. In this protocol
the seller does not get to know the exact watermarked copy
that the buyer receives.Hence seller cannot create duplicates of

EKternal data/pay

the keys
Case 3

::]
::::::::
c.:a:s:e2-------------------1t-------17f:re :c a:-v:erh:as------------

If re c: e.i ver has

encrvp

'0'

cad

If .- e.ce l ver has bo

key o n l ..,

Decryption

...

data h id i n g

kevonlv

+
wi t h hidden dot ..

original cover Image

EKternal d ata/ p a y load

Fig. 2. Separable reversible data hiding with three cases.

772

2014 International Conference on Issues and Challenges in Intelligent Computing Techniques (ICICT)

the original content containing the buyer watermark. However,

in case the seller finds an unauthorized copy, he can identify
the buyer from whom this unauthorized copy has originated.
The watermark embedding protocol is based on public key
cryptography and has little overhead in terms of the total data
communicated between the buyer and the seller. This scheme
also stated the concept of hiding the data in encrypted form of
the data. Here seller is doing data (fingerprint/Watermark in
this case) embedding while he does not know the original data
content thus invisible watermarking. The original data content
is in the encrypted form [7] .
Most o f the work on reversible data hiding focuses on the
data embedding/extracting on the plain spatial domain. But in
some applications it is important to append some additional
message within the encrypted image however data hider does
not know the original image content. And it is also expected
that the original content should be recovered without any error
after image decryption and message extraction at receiver side.
It leads to the strong security based reversible data hiding that
it should work in encryption-decryption domain.
Reference [3] offered a practical scheme satisfying
the above-mentioned requirements. A content owner encrypts
the original cover image using an encryption key and a data
hider could embed additional data into the encrypted image
using a data-hiding key although he does not know the
original content. Having an encrypted image containing
additional data a receiver first decrypts it according to the
encryption key, and then extracts the embedded data and
recovers the original image according to the data-hiding key.
In the scheme the procedure of data extraction is not separable
from the content decryption. In other words, the principal
content of original image is revealed before payload
extraction, and, if someone has the data-hiding key only but
not the encryption key he is unable to extract any information
from the encrypted image containing additional data [3].

much text can be hiding into the image. For these reasons it is
important to take these issues into consideration.
As shown in Fig.l, in separable scheme, the content owner
encrypts the original uncompressed image using an encryption
key to produce an encrypted image. Then using a data-hiding
key the data-hider compresses the least significant bits (LSB)
of the encrypted image to create some space to accommodate
the additional data. At the receiver side the data embedded can
be easily retrieved from the encrypted image containing
additional data according to the data-hiding key. Because the
data embedding only affects the LSB of the cover image;
decryption may result in an image similar to the original
version. Here at the receiver side there exists three cases as
shown in Fig.2, case one is when the receiver has only the
data- hiding key, he is able to extract the additional data even
if he does not know the image content. Case two is if he has
only the encryption key, he can decrypt the received data i.e.
encrypted image to obtain an image similar to the original
cover media, but cannot extract the embedded additional data.
Case three is if the receiver has both the keys i.e. data-hiding
key and the encryption key, he can extract the additional data
and recovers the original image without any error. This new
approach of separable reversible data hiding method consists
of three main procedures,
A.

Selecting the cover image and make it encrypted

B.

Hiding the data

C.

Data extraction/image recovery process

A. Selecting the cover image and make it encrypted

Select the appropriate cover image in which we want to add

the data. The cover image can be of any format like jpg, jpeg,
png, gif, bmp etc. Assume that the pixels of original selected
image is ranging into [0,255] is represented by 8 bits. Denote
these bits of a pixel as seven bits. That implies
hi.j.k,u

IV.

METHODOLOGY

In common there are two kinds of reversible data hiding

techniques (according to key distribution) separable reversible
data hiding technique and non-separable reversible data hiding
technique (also called as reversible data hiding).As we know in
separable reversible data hiding we are separating two
activities i.e. cover image decryption and pay load data

lPi,j.kl2U JIIDd 2

DATA CAPACITY TO EMBED IN THE IMAGE AND RELATED

PSNR
Sr.No.

Payload (In

Approximate text

PSNR

number

file size

recovered

words)

500

2
3
4

of

(in

KB)

of

image

45.8573

1000

6.41

42.9012

1500

9.61

41.1306

2000

13. 3

39.946

0 , 1 , 2. . . .... 7

(I)

(2)

pi,j,k Lbi,j,k.". 2"

=

11=0

In this encryption phase, the exclusive-or results of the

original bits and pseudo-random bits are calculated as
=

(3)

bi,}, k,lI (f) ri,}, k,lI

Where ri,},k,u are determined by an encryption key using a

standard stream cipher.

after

decryption(in dB)

3.20

and

Bi,}, k,u
TABLE l.

extraction. In addition to this there are some further issues

related with existing separable data hiding method.
The domain used for data hiding in this new approach is
based on an image Encryption-Decryption domain. That is the
cover image in which the secure data is to be hide is encrypted
first or make it distorted. And then the data hider compresses
that encrypted version of the cover image to make a room for
additional data (Secured data).
The secure data which is to be hide used in this new
approach is large enough text data. Table I shows the how

B. Hiding the data

In this step of data embedding; data is converted into

binary
form
so
that
bits
of
the
data
i.e.
B(c1,I),B(c1,2),....., B(c1,8),......., B(cn,8) can be embedded into
LSB of the cover image. Here the data we embedded is text so
c1 is single character in the text and en means numbers of
characters are there in the text. Actually the characters in the
text are represented by their ASCII value and that value is
converted into binary format thus we get the bits of the text.
According to a data-hiding key, the data-hider randomly
selects encrypted pixels. For each pixel least significant bits
and indicate them as Bi,},k,u where u 5,6,7,8 these are the
bits used to carry the bits of the text. Here actually it requires
two pixels to store eight bits or single character. Because we
are storing the text bits only into LSB of the image pixels that
last four bits per pixel in this case. Now replace the LSB of
=

20J4 Internationai Conference on Issues and Challenges in Intelligent Computing Techniques (ICICT)

773

randomly selected encrypted pixels with the bits of the

text B(el,I),B(el,2),....., B(el,8), ......., B(cn,8) .
The embedding rate, a ratio between the data amount of
net payload and the total number of cover pixels, is
ER=

total no characters in the text(payloadj

total_no_of _cover _pixels

(4)

C. Data extraction/image recovery process

In this step, we will consider the three cases at the receiver

side that a receiver has the data-hiding key only, the encryption
key only, and both the data-hiding as well as encryption keys
respectively.
Case one is with an encrypted image containing the
embedded data if the receiver knows only a data-hiding key, he
first get the values of the LSB of the encrypted image
containing secure text. Now these bits contains the bits of the
text so extract the two group of LSB of the image pixel each
having four bits then make a group of eight bits i.e. a byte then
convert that byte into equivalent value since this is the ASCII
value of the character of the text. Finally transfer that value
into equivalent character so that we get the complete text as a
string.
Note that since the random pixel selection, any attacker
without the knowledge data-hiding key cannot obtain the
values and the statistics about pixel-groups, for that reason he
cannot extract the embedded data. Furthermore, while the
receiver having the data-hiding key can successfully extract the
embedded data, he is unable get any information about the
original image content.
Consider the second case in which the receiver has the
encryption key but does not know the data-hiding key. Clearly,
he cannot extract the embedded data. But, the original image
content can be roughly recovered at the receiver. Denoting the
bits of pixels in the image containing embedded data as
B'i,j,k,O, B'i,j.k.I,.B'i,j,k,2,......... ..... B'i,j,k,7 the receiver can decrypt
the received data
h'i,j,k,l/

B'i,j,k,l/ iCBri,j,k,l/

(5)

Since the data-embedding operation does not alter any

MSB of encrypted image, the decrypted MSB must be same as
the original MSB. So the content of decrypted image is similar
to that of original cover image. The Mean Square Error for that
decrypted image is

V.

EVALUATION METHODOLOGY

In this new approach we are concentrating on how the

performance of the current scheme can be increased by
considering few issues like
A.Increase the amount of data to be hide

The present scope of the technique is that the author in [4]

has suggested the method of separable reversible data hiding
when the amount of secure data is not too large. Size of the
data must be small. So we need such approach which will
overcome this size limitation. Here in proposed system we are
trying to embed large amount of data into encrypted version of
the cover image. By observing Table 1 it is clear that the
proposed separable scheme can handle enough data for
security.
B. Getting higher PSNR of the decrypted version of the cover
image

The proposed scheme's main advantage is that we recover

the original cover image successfully without error and having
minimum or zero distortion. Here we are interested in PSNR
of the decrypted cover image after performing encryption
decryption, data hiding and data extracting process. As shown
in Fig. 3 it is clear that there is very negligible loss in
decrypted image since we get the high PSNR values. How the
quality is maintained even after performing encryption
decryption, data hiding and data extracting process on the
image is shown in Fig. 3.

60
__ PSNR of decrypted image
50

<C
...
;;:

40

c:

:z
(I)
c..

30

20

10

0
0

0.05

0.1

0 .15

0.2

0.25

Embedded rate

(6)

Fig. 3. Achieving higher PSNR of decrypted version of the cover image.

1.005

Where D is the difference of original cover image contains

and
the
decrypted
cover
image
contains.
So, the value of PSNR in the directly decrypted image is
PSNR=20.10g12551 MSE)

0.995

(7)

Consider the third case in which the receiver has both the
data-hiding and the encryption keys, he may aim to extract the
embedded data i.e. text and recover the original image. Here
according to the data-hiding key the receiver first get the values
of the LSB of the encrypted image containing secure text since
the further procedure is same as in case one. Similar to the case
two if the receiver has encryption key he can easily get the
original cover image.

.Orange.jpg

..

0.99

.S:

Leaves. jpg
.....
-

a 0.985
"

..,..

-Lena.jpg

0.98

;; 0.975

..
a

0.97
0.965
0

0.02 0 .040.06 0.08 0.1 0.12 0.140.16 0.18 0.2 0.22 0.24
embedded rate

Fig. 4. Observing high quality index.

774

2014 International Conference on Issues and Challenges in Intelligent Computing Techniques (ICICT)

C.

scheme. Sometime the secure data can be in any form rather

Observing high quality index.

The normal range of universal quality index is -1 to

I.Higher quality index hints to better quality. It is observed

that

we get the unit quality index

than text for example an image. So in such situation we will

have to hide an image data into cover image.

for various possible

embedded rate and for various images as shown in Fig. 4.1t

also shows how the quality index ranges from 0.98 to l.00 .
So it is concluded that the quality is not compromising in any

REFERENCES
[I ]

Z. Ni, Y.-Q. Shi, N. Ansari, and W. Su, "Reversible data hiding," IEEE
Trans. Circuits Syst. Video Techno!., vo!' 16, no. 3, pp. 354-362,
Mar.2006.

way.
[2 ]
CONCLUSION

The main objective of this study is to describe the concept

of separable reversible data hiding using new approach. This
new approach describes how we can maintain the performance
after increasing the amount of payload. So after studying this

Aug. 2003.
[3 ]

X. Zhang, "Reversible data hiding in encrypted image," IEEE Signal

[4 ]

X Zhang, "Separable Reversible Data Hiding in Encrypted Image",

pp. 826-832,Apr.2012.

[5 ]

as well as quality of the cover image. In this scheme we are not

degrading the quality of the cover media as we get the unit
quality index of original cover image and the image after

encryption-decryption, data hiding and data extracting process

M. Johnson, P. Ishwar, V. M. Prabhakaran, D. Schonberg, and K.

Ramchandran, "On compressing encrypted data," IEEE Trans. Signal
Process., vo!. 52, no. 10, pp. 2992-3006, Oct. 2004.

[6 ]

encryption-decryption, data hiding process. Also higher PSNR

of the decrypted cover image is observed after performing

Process. Lett., vo!. 18, no. 4, pp. 255-258, Apr. 20I I.

IEEE transactions on information forensics and security, vo!. 7, no. 2,

novel technique it has been concluded that it is possible to hide

enough or large amount of data without compromising security

J. Tian, "Reversible data embedding using a difference expansion,"

IEEE Trans. Circuits Syst. Video Techno!., vo!' 13, no. 8, pp. 890-896,

W. Liu, W. Zeng, L. Dong, and Q. Yao, "Efficient compression of

encrypted grayscale images," IEEE Trans. Image Process., vol. 19, no.
4,pp. 1097-1102, Apr. 2010.

[7]

N. Memon and P. W. Wong, "A buyer-seller watermarking protocol,"

IEEE Trans. Image Process., vo!. 10, no. 4, pp. 643-649, Apr. 2001.

on cover image. Still there can be more future aspects in this

20J4 Internationai Conference on Issues and Challenges in Intelligent Computing Techniques (ICICT)

775