Reversible encrypted data concealment in images by reserving room approach

Proceedings of the International Conference on Emerging Trends in Engineering and Management (ICETEM14)
30 – 31, December 2014, Ernakulam, India
203
REVERSIBLE ENCRYPTED DATA CONCEALMENT IN
IMAGES BY RESERVING ROOM APPROACH
MINNU T UMMER1
, KAVITHA N NAIR2
1
(PG Scholar, Department of ECE, University College of Engineering, Muttom, Kerala)
2
(Lecturer, Department of ECE, University College of Engineering, Muttom, Kerala)
ABSTRACT
Recently, more and more attention is paid to reversible data hiding (RDH) in encrypted images, since it
maintains the excellent property that the original cover can be losslessly recovered after embedded data is extracted while
protecting the image content’s confidentiality. All previous methods embed data by reversibly vacating room from the
encrypted images, which may be subject to some errors on data extraction and/or image restoration. Here propose a novel
method by reserving room before encryption with a traditional RDH algorithm, and thus it is easy for the data hider to
reversibly embed data in the encrypted image. The proposed method of reserving room before encryption is based on the
image decomposition by using Lifting Wavelet Transform (LWT) and this is compared with the traditional method to
obtain the better results. The proposed method is free of any error. Experiments are done on five test images to show that
this novel method can embed more than 10 times as large payloads for the same image quality as the previous methods,
such as for PSNR 40 dB and to show the method based on LWT can achieve better performance in terms of PSNR and
MSE.
Keywords: Image Encryption, LSB Embedding, Privacy Protection, Performance Analysis, Reversible Data Hiding.
1. INTRODUCTION
Reversible data hiding in encrypted images is a new topic getting attention because of the secured
environmental requirements. Data hiding in reversible manner in encrypted images is providing double security for the
data such as image encryption as well as data hiding in encrypted images. Reversible data hiding (RDH) in images is a
technique, by which the original cover can be losslessly recovered after the embedded message is extracted. The
reversibility means not only embedding data but also original image can be precisely recovered in the extracting stage.
However in a number of domains such as military, legal and medical imaging where no distortion of the original cover is
allowed, this highlights the need for Reversible (Lossless) data embedding techniques. In applications such as in law
enforcement, medical images systems, it is desired to be able to reverse the stego media back to the original cover media
for legal consideration. The remote sensing and military imaging, high accuracy is required. In some scientific research,
experimental data are expensive to be achieved. Under these circumstances, the reversibility of the original media is
desired. In practical aspect, many RDH techniques have emerged in recent years. By first extracting compressible
features of original cover and then compressing them losslessly, spare space can be saved for embedding auxiliary data.
In theoretical aspect, Kalker and Willems [1] established a rate-distortion model for RDH, through which they
proved the rate-distortion bounds of RDH for memoryless covers and proposed a recursive code construction which,
however, does not approach the bound. Zhang et al. [2], improved the recursive code construction for binary covers and
proved that this construction can achieve the rate-distortion bound. In practical aspect, many RDH techniques have
INTERNATIONAL JOURNAL OF ELECTRONICS AND
COMMUNICATION ENGINEERING & TECHNOLOGY (IJECET)
ISSN 0976 – 6464(Print)
ISSN 0976 – 6472(Online)
Volume 5, Issue 12, December (2014), pp. 203-215
© IAEME: http://www.iaeme.com/IJECET.asp
Journal Impact Factor (2014): 7.2836 (Calculated by GISI)
www.jifactor.com
IJECET
© I A E M E

204
emerged in recent years. Fridrich et al. [3] constructed a general framework for RDH. By first extracting compressible
features of original cover and then compressing them losslessly, spare space can be saved for embedding auxiliary data.
A more popular method is based on difference expansion (DE) [4], in which the difference of each pixel group is
expanded, e.g., multiplied by 2, and thus the least significant bits (LSBs) of the difference are all-zero and can be used
for embedding messages. Another methods [5] usually combined DE or histogram shift (HS) to residuals of the image,
e.g., the predicted errors, to achieve better performance.
In [6], Zhang divided the encrypted image into several blocks. By flipping 3 LSBs of the half of pixels in each
block, room can be vacated for the embedded bit. The data extraction and image recovery proceed by finding which part
has been flipped in one block. This process can be realized with the help of spatial correlation in decrypted image.
Zhang’s method in [7] pseudo-randomly permuted and divided encrypted image into a number of groups with size of The
LSB-planes of each group are compressed with a parity-check matrix and the vacated room is used to embed data.
Fig. 1: Framework: “vacating room after encryption (VRAE)” versus framework: “reserving room before encryption
(RRBE). (a) Framework VRAE. (b) Framework RRBE
We elaborate a practical method based on the Framework “RRBE”, shown in Fig(1) which primarily consists of
four stages: generation of encrypted image, data hiding in encrypted image, data extraction and image recovery. Note
that the reserving operation adopt in the proposed method is a traditional RDH approach. As shown in Fig. 1(b), the
content owner first reserves enough space on original image and then convert the image into its encrypted version with
the encryption key. Now, the data embedding process in encrypted images is inherently reversible for the data hider only
needs to accommodate data into the spare space previous emptied out. The data extraction and image recovery are
identical to that of Framework VRAE. Obviously, standard RDH [8],[9] algorithms are the ideal operator for reserving
room before encryption and can be easily applied to Framework RRBE to achieve better performance compared with
techniques from Framework VRAE. This is because in this new framework, we follow the customary idea that first
losslessly compresses the redundant image content (e.g., using excellent RDH techniques) and then encrypts it with
respect to protecting privacy. Two RRBE Methods are considered here.
2. METHODOLOGY
In this section two methods of reserving room before encryption is discussed.
2.1 RRBE using LWT
The Methodologies followed in this method is (1) Lifting Wavelet Transformer, (2) Chaos based image
encryption, (3) Adaptive LSB Replacement, (4)Data Recovery by decryption and is shown in fig. 2.

Fig 2: Block Diagram of image encryption and data hiding
2.1.1 Lifting wavelet transformer (LWT)
Wavelet transform [10] provides time frequency representation. The wavelet transform of an image is created by
repeated filtering the image coefficients on a row by row and column by column basis. To generate four wavelet bands
such as Approximation band, Vertical band, Diagonal band and Horizontal band. The approximation band has low
frequency component and significant information is present i
insignificant information of an image such as detailed and minute information such as edge information, corner detailed
information etc., Daubechies (db2) wavelet transform is applied on cover im
wavelet (Daubechies db1) is used to transform payload into wavelet domain. Harr wavelet is discontinuous and
resembles a step function which represents the same wavelet as Daubechies (db1).
The lifting scheme (LS) [10] has been introduced for the efficient computation of Discrete Wavelet
Transform(DWT)[10].For image compression, it is very necessary that the selection of transform should reduce the size
of the resultant data as compared to the original data s
Wavelet using the lifting scheme significantly reduces the computation time, speed up the computation process. The
lifting transform even at its highest level is very simple. The lifting transform
Predict and Update. Suppose we have the one dimensional signal a0. Lifting is done by performing the following
sequence of operations:
1. Split a0 into Even-1 and Odd-1
2. dj-1 = Oddj-1 – Predict (Evenj-1)
3. aj-1 = Evenj-1 + Update( dj-1 )
These steps are repeated to construct multiple scales of the transform. The wire diagram in Fig. 3 shows the
forward transform visually. The coefficients “a” are representing the averages in the signal that is Approximation
coefficient, while the coefficients in “d” represent the differences in the signal that is Detailed Coefficient. Thus, these
two sets also correspond to the low- pass and high
Fig 3: Wire diagram of forward tran
International Conference on Emerging Trends in Engineering and Management (ICETEM14)
205
Block Diagram of image encryption and data hiding
(LWT)
oefficients on a row by row and column by column basis. To generate four wavelet bands
frequency component and significant information is present in this band. The vertical, diagonal and horizontal bands has
information etc., Daubechies (db2) wavelet transform is applied on cover image to convert into wavelet domain and Haar
resembles a step function which represents the same wavelet as Daubechies (db1).
e (LS) [10] has been introduced for the efficient computation of Discrete Wavelet
of the resultant data as compared to the original data set .So a new lossless image compression method is proposed.
lifting transform even at its highest level is very simple. The lifting transform can be performed via two operations: Split,
1
1)
pass and high- pass frequencies present in the signal.
Wire diagram of forward transformation with the lifting scheme
31, December 2014, Ernakulam, India
oefficients on a row by row and column by column basis. To generate four wavelet bands
n this band. The vertical, diagonal and horizontal bands has
age to convert into wavelet domain and Haar
e (LS) [10] has been introduced for the efficient computation of Discrete Wavelet
et .So a new lossless image compression method is proposed.
can be performed via two operations: Split,
sformation with the lifting scheme

206
The inverse transformation is also very simple as well. We only reverse the order of operations and change the
signs. The even and odd sequences are then merged together to form the original signal. The wire diagram of inverse
transformation is shown below in Fig. 4
s
d
Fig 4: Wire diagram of Inverse Transformation with the lifting scheme
LWT decomposes the image into different subband images, shown in Fig 5. namely, LL, LH, HL, and HH for
embedding the messages in the pixel coefficients of subbands. Lifting scheme is a technique to convert DWT coefficients
to Integer coefficients without losing information. LL subbands contains the significant part of the spatial domain image.
High-frequency subband contains the edge information of input image. These coefficients are selected as reserved space
for hiding the text data. The secret text data is embedded into the wavelet coefficients of high frequency subbands
because it is non sensitive to human visual system.
Forward Lifting in IWT is calculated by following steps:
Step1: Column wise processing to get H and L
H = (Co-Ce) and L = (Ce+ [H/2]) (1)
Where Co and Ce is the odd column and even column wise pixel values.
Step 2: Row wise processing to get LL,LH,HL and HH. Separate odd and even rows of H and L,Namely, Hodd – odd
row of H, Lodd- odd row of L,Heven- even row of H, Leven- even row of L.
LH = Lodd-Leven (2)
LL = Leven + [LH / 2] (3)
HH = Hodd – Heven (4)
HL = Heven + [HH / 2] (5)
Fig 5: Block Diagram of LWT
Update Predict Merge
-
+
a(n)

Reverse Lifting scheme in IWT
Procedure is similar to the forward lifting scheme.
cover image and transformed image is shown in Fig.6.
Fig 6
2.1.2 Chaos Encryption
Chaos is a dynamical system that is extremely sensitive to its initial conditions. It is a deterministic nonlinear
system that has random-like behaviors. Chaos theory has become a new branch of scientific studies today. Discrete
chaotic dynamic systems are used in this system. The implemented map is logistic map,[11],[12] which is one of the
simplest form of one dimensional chaotic maps and mathematically its equation (6) can be written as:
Xn+1 = u*x(1-x)
Where x is a real value in (0,1), and u is bifurcation parameter satisfying 0
initial value x0 represents the key. The logistic map is chaotic on the condition 0.3
advanced encryption standard to encrypt the image for secure transmission.
with encryption key value generated from chaotic sequence with threshold function by bitxor operation
is used for generation of chaotic map sequence.
securely which prevents data hacking. The flow diagram is shown in Fig.7.
Fig .7
2.1.3 Adaptive LSB Embedding
A 8-bit gray scale image matrix consisting m × n pixels and a secret message consisting of k bits.
message is embedded into the LSB of the first pixel and the second bit of message is embedded into the
so on.The resultant Stego-image which holds the secret message is also a 8
207
Reverse Lifting scheme in IWT: Inverse Integer wavelet transform is formed by Reverse lifting scheme.
e is similar to the forward lifting scheme. Inverse wavelet transform is important to get the original image.The
cover image and transformed image is shown in Fig.6.
Fig 6: Cover Image and Transformed Image
a dynamical system that is extremely sensitive to its initial conditions. It is a deterministic nonlinear
like behaviors. Chaos theory has become a new branch of scientific studies today. Discrete
this system. The implemented map is logistic map,[11],[12] which is one of the
(6)
Where x is a real value in (0,1), and u is bifurcation parameter satisfying 0≤ u ≤4.n=0,1,.......The parameter U and the
represents the key. The logistic map is chaotic on the condition 0.35699≤ u ≤4.
advanced encryption standard to encrypt the image for secure transmission. It encrypts the original image pixel values
with encryption key value generated from chaotic sequence with threshold function by bitxor operation
is used for generation of chaotic map sequence. It is very useful to transmit the secret image through unsecure channel
The flow diagram is shown in Fig.7.
Fig .7: Flow diagram for chaotic Encryption
bit gray scale image matrix consisting m × n pixels and a secret message consisting of k bits.
message is embedded into the LSB of the first pixel and the second bit of message is embedded into the
image which holds the secret message is also a 8-bit gray scale image and difference between
Inverse Integer wavelet transform is formed by Reverse lifting scheme.
Inverse wavelet transform is important to get the original image.The
a dynamical system that is extremely sensitive to its initial conditions. It is a deterministic nonlinear
like behaviors. Chaos theory has become a new branch of scientific studies today. Discrete
this system. The implemented map is logistic map,[11],[12] which is one of the
4.n=0,1,.......The parameter U and the
≤4. This method is one of the
It encrypts the original image pixel values
with encryption key value generated from chaotic sequence with threshold function by bitxor operation Here logistic map
It is very useful to transmit the secret image through unsecure channel
bit gray scale image matrix consisting m × n pixels and a secret message consisting of k bits. The first bit of
message is embedded into the LSB of the first pixel and the second bit of message is embedded into the second pixel and
bit gray scale image and difference between

the cover image and the Stegoimage is not visually perceptible.
increase in number of LSBs. This hiding process will introduce the error between input and output image and it is
determined by mean square error and Peak signal to noise ratio determines the image quality.
shown in Fig 8.
2.1.4 Data Extraction and Image Restoration
Since data extraction is completely independent from image decryption, the order of them implies two different
practical applications. To manage and update personal information of images
privacy, an inferior database manager may only get access to the data hiding key and have to manipulate data in
encrypted domain. The order of data extraction before image decryption guarantees the feasibility o
case. When the database manager gets the data hiding key, he can decrypt the LSB
data by directly reading the decrypted version. When requesting for updating information of encrypted images, the
database manager, then, updates information through LSB replacement and encrypts updated information according to
the data hiding key all over again. As the whole process is entirely operated on encrypted domain, it avoids the leakage
of original content. Retrivel [13]of data and the image consist of,
1. Decompose the stego image into four bands using Daubechies
2. Detail CH band is used for extracting payload
3. Using the extracting Function payload is extracted by retrieving the l
4. Then chaotic decryption is used for cover image retrieval.
5. Then perform the inverse LWT to get the cover image.
2.2 RRBE Without Using LWT.(Traditional Method)
This Method consists of five steps
(4) Image decryption, (5) Data extraction and Image recovery.
2.2.1. Image Partition
To construct the encrypted image, the very first stage is being divided into three steps: image partition,
reversible embedding [14] followed by image encryption. Initially, image partition step divides original image into two
parts A and B then, the LSBs of A are reversibly embedded into B with a standard RDH algorithm so that LSBs of A can
be used for accommodating messages; at last, encrypt the rearranged image to generate its final version. The operator
here for reserving room before encryption is a standard RDH technique, so the goal of image partition is to construct a
smoother area B , on which standard RDH algorithms can achieve better performance. To do that, without loss of
generality, assume the original image C is an 8 bits gray
1≤ i ≤ M, 1≤ j ≤ N. First, the content owner extracts from the original image, along the rows
whose number is determined by the size of to
rows, where m = [l/N] and the number of blocks can be computed through n = M
208
the cover image and the Stegoimage is not visually perceptible. The quality of the image, however degrades with the
This hiding process will introduce the error between input and output image and it is
determined by mean square error and Peak signal to noise ratio determines the image quality.
Fig.8: LSB embedding block diagram
2.1.4 Data Extraction and Image Restoration
To manage and update personal information of images which are encrypted for protecting clients’
encrypted domain. The order of data extraction before image decryption guarantees the feasibility o
case. When the database manager gets the data hiding key, he can decrypt the LSB-planes of and extract the additional
ase manager, then, updates information through LSB replacement and encrypts updated information according to
vel [13]of data and the image consist of,
to four bands using Daubechies Lifting Wavelet Transformation.
Detail CH band is used for extracting payload.
payload is extracted by retrieving the least two bits of CH band of stego image.
Then chaotic decryption is used for cover image retrieval.
Then perform the inverse LWT to get the cover image.
2.2 RRBE Without Using LWT.(Traditional Method)
This Method consists of five steps. (1) Image Partition, (2) Self Reversible embedding,
(5) Data extraction and Image recovery.
To construct the encrypted image, the very first stage is being divided into three steps: image partition,
commodating messages; at last, encrypt the rearranged image to generate its final version. The operator
ard RDH algorithms can achieve better performance. To do that, without loss of
C is an 8 bits gray-scale image with its size M x N and pixels Cij
N. First, the content owner extracts from the original image, along the rows,
whose number is determined by the size of to-be-embedded messages, denoted by l. In detail
/N] and the number of blocks can be computed through n = M – m + 1. An important
The quality of the image, however degrades with the
This hiding process will introduce the error between input and output image and it is
determined by mean square error and Peak signal to noise ratio determines the image quality. The block diagram is
which are encrypted for protecting clients’
encrypted domain. The order of data extraction before image decryption guarantees the feasibility of our work in this
planes of and extract the additional
ase manager, then, updates information through LSB replacement and encrypts updated information according to
Lifting Wavelet Transformation.
east two bits of CH band of stego image.
(2) Self Reversible embedding, (3) Image Encryption,
To construct the encrypted image, the very first stage is being divided into three steps: image partition, self
commodating messages; at last, encrypt the rearranged image to generate its final version. The operator
ard RDH algorithms can achieve better performance. To do that, without loss of
ith its size M x N and pixels Cij€ [0, 255],
, several overlapping blocks
. In detail, every block consists of
. An important point here is that

209
each block is overlapped by pervious and/or sub-sequential blocks along the rows. For each block, define a function to
measure its first-order smoothness.
f=∑ ∑ ‫ܥ‬௨,௩
ேିଵ
௩ୀଶ
௠
௨ୀଶ −
஼ೠషభ,ೡశ಴ೠశభ,ೡశ ಴ೠ,ೡషభశ಴ೠ,ೡశభ
ସ
(7)
Higher f relates to blocks which contain relatively more complex textures. The content owner, therefore, selects
the particular block with the highest to f be A, and puts it to the front of the image concatenated by the rest part B with
fewer textured areas, as shown in Fig. 9. It is obvious that the content owner can also embed two or more LSB-planes of
A into B, which leads to half, or more than half, reduction in size of A.
Fig 9: Illustration of Image partition and embedding process
However, the performance of A decreases significantly in terms of PSNR, after embedding the data in the
second stage with growing bit-planes exploited. Hence, we investigate situations that at most three LSB-planes of A are
employed and determine the number of bit-plane with regard to different payloads.
2.2.2 Self-Reversible Embedding
The motive of self-reversible embedding [14] is to embed the LSB-planes of A into B by employing traditional
RDH algorithms. Pixels in image B are first categorized into two sets as, white pixels with its indices i and j satisfying ( i
+j)mod 2=0 and black pixels with indices ( i +j)mod 2= 1 as in Fig. 9. Then, each white pixel Bi,j is estimated by the
interpolation value obtained with the four black pixels surrounding it as follows,
B୧,୨
′
= wଵB୧ିଵ,୨ + wଶB୧ାଵ,୨ + wଷB୧,୨ିଵ + wସB୧,୨ାଵ (8)
Where the weight wi, 1 ≤ i ≤ 4, Then the estimating error is calculated via eij = Bi,j – B’i,j along with
embedding some data into the estimating error sequence with histogram shift. Then, we further calculate the estimating
errors of black pixels with the help of surrounding white pixels that may have been modified. Then another estimating
error sequence is produced that can accommodate messages. Thus we summarize that, to exploit all pixels of B, two
estimating error sequences are constructed for embedding messages in every single-layer of embedding process.
Using bidirectional histogram shift, some messages can be embedded on each error sequence i.e. firstly we
divide the histogram of estimating errors into two parts namely the left part and the right part, and search for the highest
point in each part, denoted by LM and RM, respectively. For typical images, LM = -1and RM=0. Further, look for the
zero point in each part, denoted by LN and RN. To embed messages into positions with an estimating error that is equal
to RM , shift all error values between RM+1 and RN-1 with one step towards right, and then, we can represent the bit 0
with RM and the bit 1with RM=1. The embedding process in the left part is similar except that the shifting direction is
left, and the shift is realized by subtracting 1 from the corresponding pixel values.
In RDH algorithms, there occurs the overflow and underflow problem when the natural boundary pixels change
from 255to 256. For its avoidance, just embed data into estimating error with its corresponding pixel that are valued from
1 to 254. However, problems still arise when non-boundary pixels are changed from 1 to 0 or from 254 to 255 during the
embedding process. These created boundary pixels are defined as pseudo-boundary pixels in the embedding process.
Hence, a boundary map is introduced to indicate whether boundary pixels in marked image are natural or pseudo in
extracting process.

210
2.2.3. Image Encryption
After the rearranged self-embedded image which is denoted by X is generated, we encrypt X to construct the
encrypted image denoted by E .Using stream cipher; the encryption version of X can be easily obtained. For example,
agray value Xi,j ranging from 0 to 25 can be represented by8 bits, Xi,j(0), Xi,j(1), . . . , Xi,j(7), such that,
ܺ௜,௝ሺ݇ሻ = ቂ
௑೔,ೕ
ଶೖ ቃ ݉‫,2݀݋‬ ݇ = 0,1, … .7 (9)
The encrypted bits Ei,j(k) can be calculated through exclusive-or operation.
‫ܧ‬௜,௝ሺ݇ሻ=ܺ௜,௝ሺ݇ሻ ⊕ ‫ݎ‬௜,௝ሺ݇ሻ (10)
Where ri,j(k) is generated via a standard stream cipher determined by the encryption key. Finally, we embed 10
bits information into LSBs of first 10 pixels in encrypted version of A to indicate data hider the total number of rows and
the bit-planes he can embed information into. Since after image encryption, none of the data hider and the third party
access the content of original image without the encryption key, hence privacy of the content owner is protected.
2.2.4 Image Decryption
With the encryption key, the content owner decrypts the image except the LSB-planes of AE. The decrypted
version of E' containing the embedded data can be calculated by
ܺ௜,௝
"
ሺ݇ሻ = ‫ܧ‬௜,௝
′
ሺ݇ሻ ⊕ ‫ݎ‬௜,௝ሺ݇ሻ (11)
ܺ௜,௝
"
= ∑ ܺ௜,௝
"଻
௞ୀ଴ ሺ݇ሻ × 2௞
(12)
Where E’i,j(k) and X”i,j(k) are the binary bits of E’i.j and X”i,j obtained via (11) and (12) respectively.
2.2.5 Data Extraction and Image Recovery
The content owner can further extract the data and recover original image after generating the marked decrypted
image. The process is similar to the traditional RDH methods. The following outlines the specific steps[14]:
• Step 1. Record and decrypt the LSB-planes of A” according to the data hiding key; extract the data until the end
label is reached.
• Step 2. Extract LN, RN, LM, RM, LP, RP, Rb, x and boundary map from the LSB of marginal area of B”. Then,
scan B” to -undertake the following steps.
• Step 3. If Rb is equal to 0, which means no black pixels participate in embedding process, go to Step5.
• Step 4. Calculate estimating errors e’i,j of the black pixels B”i,j. If B”i,j belongs to [1, 254], recover the estimating
error and original pixel value in a reverse order and extract embedded bits when e’i,j is equal to LN, LM (or LP ),
RM (or RP ) and RN. Else, if B”i,j € { 0, 255 } , refer to the corresponding bit b in boundary map. If b = 0, skip this
one, else operate like B”i,j € [1, 254] . Repeat this step until the part of payload Rb is extracted. If extracted bits are
LSBs of pixels in marginal area then it restores them immediately.
• Step 5. Calculate estimating errors e’i,j of the white pixels B”i,j ,and extract embedded bits and recover white pixels
in the same manner with Step 4. If extracted bits are LSBs of pixels in marginal area, restore them immediately.
• Step 6. Continue doing Step 2 to Step 5 x - 1 rounds on B” and merge all extracted bits to form LSB-planes of A.
Until now, we have perfectly recover B.
• Step 7. Replace marked LSB-planes of A” with its original bits extracted from B” to get original cover image C.
2.3 Performance Analysis
The quality of marked decrypted images is compared in the terms of PSNR and MSE [13].The performance can
be measured by these two quantities.
Mean Square Error (MSE): It is defined as the square of error between cover image and stego image. The distortion in
the image can be measured using MSE.
Peak Signal to Noise Ratio (PSNR): It is the measure of quality of stego image as compared to cover image, i.e., the
percentage of noise present in the cover image.

211
3. RESULT AND DISCUSSION
The Table I and II shows the values Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR) for different
images for different image formats of the Proposed method and traditional method respectively.
The test images are gray scale images of Baboon.png, Airplane.png, Environment.jpg, Fruits.png, Nature.jpg.
are shown in Fig.10. The performance is measured interms of PSNR and MSE..The cover image of size 256*256 is used
and the payload is of 53 bytes. The data embedded in this work is,” calicut university institute of engineering and
technology@”.
(a) (b) (c) (d) (e)
Fig.11: Test Images (a) Airplane (b) Baboon (c) Environment (d) Fruits (e) Nature
The value of PSNR and MSE should varies from payload of different sizes and it also varies for different cover
image sizes. The quality of the image degrades when size of the payload increases.
Table I: PSNR & MSE Values Of Test Images Using LWT
Input Images PSNR MSE
Nature 82.1459 0.0010
Environment 73.3953 0.003
Baboon 76.6108 0.0014
Airplane 74.9284 0.0021
Fruits 76.4279 0.0015
Table II: PSNR & MSE Values Of Test Images Of Traditional Method
Input Images PSNR MSE
Nature 48.05 1.02
Environment 48.74 0.87
Baboon 34.83 2.14
Airplane 47.59 1.13
Fruits 46.64 1.41
From the above two tables we can see that the PSNR value is improved in the case of test images using LWT
decomposition. The Fig. 12, 13, 14, 15. shows the graphical representation of PSNR and MSE values of reserving room
before encryption by using LWT and without using LWT .

212
From the graphical representation, we can see that nature image has high PSNR value and lowest MSE value.
Thus we can conclude that the image decomposition method using LWT is better than the previous method by measuring
the MSE and PSNR of two proposed methods.
Fig.12: Performance interms of PSNR (db) using LWT Fig.13: Performance in terms of MSE Using LWT
Fig 14: Performance interms of PSNR in RRBE Fig 15: Performance interms of MSE in RRBE
without LWT without LWT
3.1 Comparisons and Results
We take a image of nature shown in Fig.16: Resultant Image; to demonstrate the feasibility of proposed method
using LWT image decomposition.

213
(a) (b) (c)
(d) (e)
Fig 16: Resultant Image (a) Cover image (b) LWT transformed Image (c) Stego Image (d) Encrypted Image
(e) Recovered Cover Image
The proposed method is compared with the existing method for the different images given above. The proposed
method have a significant improvement in the image quality over the existing RRBE and evaluated in terms of
performance parameters like PSNR and MSE. The existing method maybe introduce some errors on data extraction
and/or image restoration, while the proposed method is free of any error for all kinds of images. The two graphs shown in
Fig.17 and 18 give the variation of PSNR & MSE of the modified proposed method and the existing method.
Fig 17: Plot of MSE between current method and modified method

214
Fig 18: Plot of PSNR between current method and modified method
It can be clearly seen from above two graphs that the modified method have the improved PSNR and MSE values.
4. CONCLUSION
Reversible data hiding in encrypted images is a new topic drawing attention because of the privacy-preserving
requirements from cloud data management. Here performance comparison LWT based RRBE and traditional RRBE is
done . LWT based RRBE outperforms the other method. Performance of the system is evaluated based on PSNR and
MSE. The proposed method can take advantage of all traditional RDH techniques for plain images and achieve excellent
performance without loss of perfect secrecy. Furthermore, this novel method can achieve real reversibility, separate data
extraction and greatly improvement on the quality of marked decrypted images.
The future work of this project would be Reversible Data Hiding using color images. Also we can use audio,
video in case of image as cover for hiding the data.
REFERENCES
[1] T. Kalker and F.M.Willems, “Capacity bounds and code constructions for reversible data- hiding,” in Proc. 14th
Int. Conf. Digital Signal Processing (DSP2002), 2002, pp. 71–76
[2] W. Zhang, B. Chen, and N. Yu, “Capacity-approaching codes for reversible data hiding,” in Proc 13th
Information Hiding (IH’2011), LNCS 6958, 2011, pp. 255–269, Springer-Verlag.
[3] J. Fridrich and M. Goljan, “Lossless data embedding for all image formats,” in Proc. SPIE Proc. Photonics West,
Electronic Imaging, Security and Watermarking of Multimedia Contents, San Jose, CA, USA, Jan. 2002,
vol. 4675, pp. 572–583.
[4] J. Tian, “Reversible data embedding using a difference expansion,” IEEE Trans. Circuits Syst. Video Technol.,
vol. 13, no. 8, pp. 890–896, Aug. 2003.
[5] X. L. Li, B. Yang, and T. Y. Zeng, “Efficient reversible watermarking based on adaptive prediction-error
expansion and pixel selection,” IEEE Trans. Image Process., vol. 20, no. 12, pp. 3524–3533, Dec.2011
[6] X. Zhang, “Reversible data hiding in encrypted images,” IEEE Signal Process. Lett., vol. 18, no. 4, pp. 255–258,
Apr. 2011.
[7] X. Zhang, “Separable reversible data hiding in encrypted image,” IEEE Trans. Inf. Forensics Security, vol. 7,
no. 2, pp. 826 832, Apr. 2012.
[8] L. Luo et al., “Reversible image watermarking using interpolation technique,” IEEE Trans. Inf. Forensics
Security, vol. 5, no. 1, pp. 187–193,Mar. 2010.
[9] V. Sachnev, H. J. Kim, J. Nam, S. Suresh, and Y.-Q. Shi, “Reversible watermarking algorithm using sorting and
prediction,” IEEE Trans. Circuits Syst. Video Technol., vol. 19, no. 7, pp. 989–999, Jul. 2009.

215
[10] Mrs. Preet Kaur, Geetu lalit, “Comparative Analysis of DCT, DWT & LWT for Image Compression” (IJITEE)
ISSN: 2278-3075, Volume-1, Issue-3, August 2012.
[11] Haojiang Gao *, Yisheng Zhang, Shuyun Liang, Dequn Li,” A new chaotic algorithm for image encryption”
Huazhong University of Science and Technology, 1037 Luoyu Road, Wuhan 430074, PR China Accepted 16
August 2005.
[12] Mazhar Tayel, Hamed Shawky, Alaa El-Din Sayed Hafez,” A New Chaos Steganography Algorithm for Hiding
Multimedia Data” Electrical Engineering Department, Faculty of Engineering, Alexandria University, Feb.2012.
[13] H S Manjunatha Reddy, K B Raja,” Wavelet based Secure Steganography with Scrambled Payload”, International
Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-1, Issue-2,
July 2012.
[14] Kede Ma, Weiming Zhang,” Reversible data hiding in encrypted images by reserving room before encryption”,
IEEE transactions on information forensics and security, vol. 8, no. 3, march 2013.
[15] Rohini N. Shrikhande and Prof. Vinayak K. Bairagi, “Prediction Based Lossless Medical Image Compression”,
International Journal of Electronics and Communication Engineering & Technology (IJECET), Volume 4,
Issue 2, 2013, pp. 191 - 197, ISSN Print: 0976- 6464, ISSN Online: 0976 –6472.
[16] Vismita Nagrale, Ganesh Zambre and Aamir Agwani, “Image Stegano-Cryptography Based on LSB Insertion &
Symmetric Key Encryption”, International Journal of Electronics and Communication Engineering & Technology
(IJECET), Volume 2, Issue 1, 2011, pp. 35 - 42, ISSN Print: 0976- 6464, ISSN Online: 0976 –6472.

Reversible encrypted data concealment in images by reserving room approach

More Related Content

What's hot (20)

Similar to Reversible encrypted data concealment in images by reserving room approach (20)

More from IAEME Publication (20)

Recently uploaded (20)

Reversible encrypted data concealment in images by reserving room approach