See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/308917011

An Image Steganography Algorithm using Haar

Discrete Wavelet Transform with Advanced
Encryption System

Data October 2016

CITATIONS READS

0 94

3 authors:

Essam H Houssein Mona Abdelbaset Sadek Ali

Minia University Benha University
9 PUBLICATIONS 27 CITATIONS 10 PUBLICATIONS 10 CITATIONS

SEE PROFILE SEE PROFILE

Aboul Ella Hassanien

Cairo University
704 PUBLICATIONS 2,995 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

using multi-spectral and hyper-spectral image satellite to detection important diseases of sugar beet
in Egypt View project

Automatic Soccer Video Summarization View project

All content following this page was uploaded by Essam H Houssein on 07 October 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Proceedings of the Federated Conference on Computer Science DOI: 10.15439/2016F521
and Information Systems pp. 641644 ACSIS, Vol. 8. ISSN 2300-5963

An Image Steganography Algorithm using Haar

Discrete Wavelet Transform with Advanced
Encryption System
Essam H. Houssein 1, ,Mona A. S. Ali 2, , and Aboul Ella Hassanien3,

Faculty of Computers and Information
1
Minia University 2 Benha University 3 Cairo University

Scientific Research Group in Egypt (SRGE) http://www.egyptscience.net

AbstractThe security of data over the internet is a crucial age is used as the cover-image to conceal the secret data. The
thing specially if this data is personal or confidential. The secret data are embedded into the cover-image by modifying
transmitted data can be intercepted during its journey from the cover-image to form a stego-image. Cryptography [2],
device to another. For that reason, we are willing to develop
a simple method to secure data. Data encryption is one method [3] and steganography [4] are the two important aspects of
to secure the messages but the intruders can still try to crack communications security. Although cryptography is a primary
it, in order to overcome this, steganography has been used to method of protecting valuable information by rendering the
hide the data into a cover media(i.e. audio, image or video). message unintelligible to outsiders [3], steganography is a step
Recently steganography attracts many researchers as a hot topic. ahead by making the communication invisible. A possible for-
This paper proposes an advanced technique for encrypting data
using Advanced Encryption System (AES) and hiding the data mula of the process may be represented as: Stegomedium =
using Haar Discreet Wavelet Transform (HDWT). HDWT aims to Covermedium + Embeddedmessage + Stegokey
decrease the complexity in image steganology while providing less In this paper, an advanced technique for encrypting data is
image distortion and lesser detectability. One-forth of the image proposed using AES and hiding the data using Haar DWT
carrying the details of the image in a region and other three technique, carrying the details of the image in a region and
regions carrying a less details of the image then the cipher text is
concealed at most two Least Significant Bits (LSB) positions in the other three regions carrying a less details of the image then the
less detailed regions of the carrier image, if the message doesnt cipher text is concealed at most two LSB positions. Extensive
fit in the first LSB only it will use the second LSB. This proposed experiments show the effectiveness of the proposed method.
algorithm covers almost all type of symbols and alphabets. The results obtained also show significant improvement than
Index TermsSteganography, Cryptography, Encryption, AES the method proposed in [5]. The remainder of the paper is
Encryption, LSB, DWT
organized as follows. Section II briefly describes the related
work. In Section III, the related main knowledge is described.
I. I NTRODUCTION
Section IV, discusses the features of the proposed technique.
TEGANOGRAPHY is a data hiding technique that has
S been mainly used in information security applications. It
is similar to watermarking and cryptography techniques, but
The performance is analyzed. Experimental results are given
in Section V. Finally, Section VI concludes this paper.

these three techniques are different in some aspects. Firstly, II. R ELATED W ORK
watermarking mainly tracks illegal copies or claims of the A few steganography approaches are briefly reviewed
ownership of digital media. It is not geared for communication. here.In [6], Tiegang et al. proposed a new image encryption
Secondly, cryptography scrambles the data with the mixture of algorithm based on hyper-chaos, which uses a new image
permutation(s) and substitution(s) so that unintended receivers total shuffling matrix to shuffle the pixel positions of the
cannot perceive the processed information. However, the fact plain-image and then the states combination of hyper-chaos
that information has been embedded into a medium (i.e., is used to change the grey values of the shuffled-image. In
watermarking) and communication has been carried out (i.e., [7], Chin-Chen et al. proposed a new steganographic method
cryptography) is known to everyone, or at least it is acceptable to increase the message load in every block of the stego-image
to reveal such a fact. Finally, steganography transmits infor- while keeping the stego-image quality acceptable. In [8], B.
mation by embedding messages into innocuous looking cover T. NilanjanDey, et al, proposed a method for hiding multiple
objects, such as digital images, to conceal the very existence images in an image based on DWT and DCT. In [9], Chen
of communication. As a result, steganography is the art and Po-Yueh et al. proposed a new steganography technique which
science of data smuggling since its goal is to hide the presence embeds the secret messages using the DWT in the frequency
of communication [1]. domain to divide the image into 4 sub bands, and it will embed
Each image hiding system consists of an embedding process the secret data in the LSB of the lowest priority band and it
and an extraction process. An innocuous-looking original im- wouldnt use the low frequency sub band that holds the most

c
978-83-60810-90-3/$25.00 2016, IEEE 641
642 PROCEEDINGS OF THE FEDCSIS. GDANSK, 2016

details to preserve the quality of the image, and it have 2 and wavelet which associate with low and high pass filters
modes and 5 cases because of the demands of the capacity or orderly. Decomposition follows the manner of dividing time
quality. separability. Meanly, only half of the samples in a signal
In [10], Chiang-Lung et al. proposed method adopts the are sufficient to represent the whole signal, doubling the
complementary embedding strategy to reduce the loss of frequency separability.
statistical property of the stego-image in spatial domain. In
[11], Chi-Kwong Chan et al. proposed a data hiding method Haar wavelet operates on data by calculating the sums
by simple LSB substitution with an optimal pixel adjust- and differences of adjacent elements. This wavelet operates
ment process. The image quality of the stego-image can be first on adjacent horizontal elements and then on adjacent
greatly improved with low extra computational complexity. In vertical elements. One important feature of the Haar wavelet
[12], KokSheik et al. presented a novel Mod4 steganographic transform is that the transform is equal to its inverse. Each
method in discrete cosine transform (DCT) domain. Mod4 is transform computes the data energy in relocated to the top
a blind steganographic method. left hand corner.
In [13], M. Juneja et al., proposed a secure methods of
information security using steganography, the first method Figure 1 shows the image Lena after one Haar wavelet
embeds In the LSB of the blue components partial green transform
components of random positions in the edges of the green
component, the second method is an adaptive method that uses
the data of the MSB of the red, green and blue components
to embed the message in the random pixels across smooth
areas, the third method is a hybrid feature detection filter that
predicts the edges in noisy conditions.
In [5], Avval et al, proposed steganography technique to
embed audio into the edges of a color image, this technique
uses the chaotic map to select the random edge pixels to embed
the bits and to choose which random LSB bit location in the Fig. 1: 2D Haar Wavelet Transform Example
selected pixel. In [14], Hemalatha S. et al. proposed a method
to hide multiple secret keys and images in a color image using After each transform is performed the size of the square
integer wavelet transform (IWT). In [11], T. Garima, proposed which contains the most important information is reduced by
an approach to encrypt message using RSA 1024 algorithm a factor of 4 as seen in Figure 2.
then it will be embedded in a cover image by modifying
the LSB technique. In [15], L. S. Ahmed et al., proposed a
method of encrypting the message by a substitution cipher
then it will be embedded using LSB insertion techniques to
achieve high capacity to be embedded and security of the
steganography method. In [10], Liu Lung et al. proposed a
jpeg steganographic using complementary embedding tech-
nique, this method is achieved by dividing the quantized DCT
coefficients and the secret bits into two parts according to a
predefined partition ratio. The two parts of DCT coefficients Fig. 2: Detailed 2D Haar Wavelet Transform
are used to embed the corresponding parts of secret bits with
different embedding algorithms. Specifically, the secret bits are
embedded by subtracting one from one part of coefficients, and
B. AES encryption technique
adding one to the other part of coefficients.
The cipher takes a plaintext block size of 128 bits, or 16
III. T HE STEGANOGRAPHY
bytes. The key length can be 16, 24, or 32 bytes (128, 192,
A. 2D-Haar-wavlet Transform or 256 bits). Data block of 4 columns of 4 bytes is state
Wavelet transform has the capability to offer some key expanded to array of words. Ordering of bytes within a
information on frequency-time domain concurrently. In matrix is by column. The cipher consists of rounds, where the
this transform, time domain is passed through high-pass number of rounds depends on the key length: 10 rounds for
and low-pass filters to extract high and low frequencies a 16-byte key, 12 rounds for a 24-byte key, and so on, each
respectively. This process is repeated for a number of times transformation takes one or more 4 X 4 matrices as input and
and each time a section of the signal is drawn out. DWT produces a 4 X 4 matrix as output (the cipher text).The key
analysis splits signal into two classes (i.e. Approximation expansion function generates N + 1 round keys, each of which
and Detail) by signal decomposition for different frequency is a distinct 4 X 4 matrix. Each round key serves as one of the
bands and scales. DWT employs two function sets: scaling inputs to the Add Round Key transformation in each round.
 ESSAM H. HOUSSEIN ET AL.: AN IMAGE STEGANOGRAPHY ALGORITHM USING HAAR DISCRETE WAVELET TRANSFORM 643

C. LSB substitution technique Algorithm 1 Proposed Approach

The wide used technique of hiding messages into the image 1: input=gray scale cover image C and message to be em-
without affecting the whole image is the LSB technique that bedded M
uses the least significant bit of each pixel to embed the 2: output= StegoImage
message into it. Figure 3 shows how the pixel can be illustrated 3: S = Encrypt M using AES technique
from the most important bit to the least important bit. The 4: If Size(S) > total first three bit of LSB in C
basic concept of LSB substitution is to embed the message in 5: Then get another cover Image AC and go to step 4
the least important bit (least significant bit) of the pixel the 6: D = Apply the DWT on C
equation for the LSB is 7: Embed S in the coefficient of D
8: Inverse D
Xi = Xi Xi mod2k + mi (1) 9: obtain (K-matrix) that contains the possible non-integer

th
in this equation the Xi is the i pixel value of the stego situation (0.0, 0.25, 0.5 and 0.75)
image, Xi is the ith pixel of the image (cover image) and mi 10: Calculate inverse 2D-DWT on each block to get the stego
denotes the decimal value of the ith block in confidential data. image.
K denotes the number of LSB places. we will use or substitute
three sub bands. This improves the capacity and permits
from the pixel. In the process of extracting the message from
high capacity with no effect on the pixel value. As the
the image pixel is to copy the least significant bit directly, and
embedding process is done on the sub bands that dont
this process is showed in the following equation (2):
contain details of the image.

mi = xi mod2k (2) 3) Perform the inverse DWT on the result of step 2, by
performing this step the resulted matrix is H, some pixels
This technique is easy and fast but it have drawback if the of H are not integers ranging from 1-255 due to LSB
message have a great size, it will affect the image and it can substitution. So we obtain (K-matrix) that contains the
be noticeable. possible non-integer situation (0.0, 0.25, 0.5 and 0.75).
4) Round the matrix H to obtain the stego image E, in order
to reconstruct the secret message we will use the k-matrix
for reconstruction.
5) Send the Stego image to the receiver with the k-matrix
Fig. 3: The places of the bits in the pixel in the description file or tag with the total number of
message bits too.
IV. P ROPOSED A PPROACH
B. Extraction process
In this research the proposed approach includes two main
process: Embedding process and Extracting process. In order to extract the original image the following steps
are followed:
A. Embedding process
1) Extract the k-matrix of the file tag of E.
As seen in Algorithm 1, we started first by reading the
cover image C and the message to be embedded M . The K = {Ki,j |1 6 MF 6 n, 1 6 j 6 NF ,
(5)
cipher message S is extracted by applying the AES encryption Ki,j {00, 01, 10, 11}}
technique on message M . After performing the AES technique
on the message, it is ready to be concealed in the image. Transform all elements of k into (0.0, 0.25, 0.5 and 0.75).
Suppose that the 8-bit gray level cover image of size MC xNC 2) Obtain the H by performing DWT which is calculated as
as : H=E+K-matrix.
3) Obtain the total number of the message from the file tag
C = {xi,j |1 6 i 6 Mc , 1 6 j 6 Nc , xi,j {1, 2, 3..., 255}} of E, extract 3 bits iteratively from the LSB of the three
(3) sub bands HHH, HHL and HLH. After completing the
S is the n-bit secret message represented as: first LSB and there are more bits of the message is still
not extracted (bits extracted message size of E != 0) Get
S = {Si |1 6 i 6 n, Si {0, 1}} (4)
the remaining bits from the second LSB in the same way
From Algorithm 1, the embedding steps will be as follows: as the last step. Extracting the two LSB of the remaining
1) Apply the DWT on the cover image and get the four bits, example [A,B]
sub-bands obtained are denoted LL, HL, LH and HH. 4) After extracting the whole bits perform the inverse AES
2) Get three bits iteratively from the cipher message and on the encrypted bits to obtain the original message.
distribute only one bit from the three bits in every one
V. R ESULTS AND D ISCUSSION
LSB of the three sub bands HH, HL and LH respectively,
if the cipher message require more LSB, our algorithm The proposed approach is applied on 512x512 8-bit
will move cursor to the second LSB of each pixel in the grayscale images Jet, Boat, Baboon and Lena. The messages
644 PROCEEDINGS OF THE FEDCSIS. GDANSK, 2016

are generated randomly with size upto maximum hiding ca- Verification of the results has been done for various length
pacity. To measure the quality, a parameter is developed to of message (LM) and calculates error for all different messages
compute the quality of the image this parameter is called the proposed technique works fine show in Figure 5. When
PSNR and defined as follows: compare our technique with scheme [5]. It gets results better
than other method. Result of PSNR for other method between
P SN R = 10 log10 (2552 /M SE) (6)
75, 80 but PSNR of our method smaller than it as seen in
The root mean square error (RMSE) has been used as a Figure 4.
standard statistical metric to measure model performance in
VI. C ONCLUSION
meteorology, air quality, and climate research studies. The
mean absolute error (MAE) is another useful measure widely There are demands of the algorithms of steganography to
used in model evaluations, denotes the mean error and its the progress with the capacity or with quality, so with this method
deference between the original image and the stego image as we aimed at the capacity demand to store as high as possible
seen in Figure 5, and the equations to compute the MAE and messages in the image with reduced effect of the quality of
MSE are: the image, and made it even harder to get the message by
1 X encrypting it before storing the message in the cover image,
M SE = (a b)2 (7)
M XN with this method the message is secured and hidden, and the
Where a denotes the pixel in the original image and b denotes cover image can take much more message size to hide. As the
the pixel in the stego image, with high PSNR means a high main goal of steganography is to hide and secure the message.
quality, this means that while the dB is low this means that
R EFERENCES
the image has been modified or there is a noise or distortion.
[1] S. Katzenbeisser and F. Petitcolas, Information hiding techniques for
steganography and digital watermarking. Artech house, 2000.
[2] A. Nissar and A. Mir, Classification of steganalysis techniques: A
study, Digital Signal Processing, vol. 20, no. 6, pp. 17581770, 2010.
[3] S. Williams, Cryptography and network security: Principles and prac-
tices, 2006.
[4] M. M. Amin, M. Salleh, S. Ibrahim, M. R. Katmin, and M. Shamsud-
din, Information hiding using steganography, in Telecommunication
Technology, 2003. NCTT 2003 Proceedings. 4th National Conference
on. IEEE, 2003, pp. 2125.
[5] N. Bhardwaj and S. Agarwal, A new technique for extracting image
information beyond visibility, International Journal of Information and
Computation Technology, vol. 3, pp. 539548, 2013.
[6] T. Gao and Z. Chen, A new image encryption algorithm based on
hyper-chaos, Physics Letters A, vol. 372, no. 4, pp. 394400, 2008.
[7] C.-C. Chang, T.-S. Chen, and L.-Z. Chung, A steganographic method
based upon jpeg and quantization table modification, Information
Sciences, vol. 141, no. 1, pp. 123138, 2002.
[8] T. Bhattacharya, N. Dey, and S. Chaudhuri, A session based mul-
Fig. 4: PSNR Results tiple image hiding technique using dwt and dct, arXiv preprint
arXiv:1208.0950, 2012.
[9] P.-Y. Chen, H.-J. Lin et al., A dwt based approach for image steganogra-
phy, International Journal of Applied Science and Engineering, vol. 4,
no. 3, pp. 275290, 2006.
[10] C.-L. Liu and S.-R. Liao, High-performance jpeg steganography using
complementary embedding strategy, Pattern Recognition, vol. 41, no. 9,
pp. 29452955, 2008.
[11] C.-K. Chan and L.-M. Cheng, Hiding data in images by simple lsb
substitution, Pattern recognition, vol. 37, no. 3, pp. 469474, 2004.
[12] K. Wong, X. Qi, and K. Tanaka, A dct-based mod4 steganographic
method, Signal Processing, vol. 87, no. 6, pp. 12511263, 2007.
[13] M. Juneja and P. S. Sandhu, A new approach for information security
using an improved steganography technique, Journal of Information
Processing Systems, vol. 9, no. 3, pp. 405424, 2013.
[14] S. Hemalatha, U. D. Acharya, A. Renuka, and P. R. Kamath, A secure
and high capacity image steganography technique, Signal & Image
Processing, vol. 4, no. 1, p. 83, 2013.
[15] S. A. Laskar and K. Hemachandran, High capacity data hiding using
lsb steganography and encryption, International Journal of Database
Fig. 5: MSE and MAE Results Management Systems, vol. 4, no. 6, p. 57, 2012.

View publication stats