D010312230

IOSR Journal of Electronics and Communication Engineering (IOSR-JECE)
e-ISSN: 2278-2834,p- ISSN: 2278-8735.Volume 10, Issue 3, Ver. I (May - Jun.2015), PP 22-30
www.iosrjournals.org
DOI: 10.9790/2834-10312230 www.iosrjournals.org 22 | Page
A Design of Novel Algorithm for Image Steganography Using
Discrete Wavelet Transformation on Beagle Board-XM
Nagarjun B1
, Dr. Manjunath H V2
1
M.Tech Student, Department of Electronics & Communication, Dayananda Sagar Institutions, India
2
Professor, Department of Electronics & Communication, Dayananda Sagar Institutions, India
Abstract :In today’s modern internet era, Security is one of the important issue in the communication and
storage of images. In order to provide security for these images, a wide variety of techniques and proposals
have been developed all across the world. The most prominent technique among all those security schemes is
the Image steganography. It is simply defined as the art or science in which the image sender embeds some
secret information into the image and sends it across a channel to the receiver. On receiving the image at the
receiver end the information that is hidden inside the image is retrieved back. In this research, we have used
discrete wavelength transformation and modified AES techniques for the process of steganography. The wavelet
is applied to the cover image in order to produce four different sub bands of different frequencies like LL, LH,
HL and HH. The secret information is encrypted with the use of modified AES technique and it is hidden inside
the LH, HL and HH sub bands of the cover image. This novel algorithm for image steganography gives better
quality images and more security compared to other conventional methods that are present today.
Implementation of this steganography process is done using Beagle Board-XM with Open CV platform. Since
we are using Open CV and Beagle Board-XM for the purpose of implementation, the delay in processing of the
image, cost and resources required will be greatly reduced.
Keywords - Image Steganography, Discrete wavelet transformation, Biorthogonal DWT, AES, Encryption,
Beagle Board-XM, Open CV, PSNR
I. Introduction
Since the inception of internet the security of information is the most vital factor in information
technology and communication. Many methods like cryptography, watermarking and encryption and decryption
techniques were developed in order to secure the information during communication. Unfortunately it was not
enough to protect the contents of the secret message from outside phishers and hackers. There was a need of a
new technique which can keep the existence of the message secret. The technique used to implement this is
called as Steganography.Coming to the history of Steganography, the Greek historian Herodotus writes in his
literary work “Histories” about a nobleman, Histaeus, who wanted to communicate with his son in law in
Greece. To communicate secretly with his son in law, he shaves one of his trusted slaves head and tattooed the
message on his scalp. After few months when the hair was grew on his scalp the slave was sent to Greece to
dispatch the secret message on his scalp [2]. As evidence, during world war the Germans developed a special
technique called as “Microdot”. Information such as important images and photographs were reduced in size
until it was a size of a sized period and were sent with a normal cover message over an insecure channel [3].
The major difference between Steganography and Cryptography is, Cryptography focuses only on
keeping the secret message or information and it is practically detectable. But in Steganography, it focuses on
keeping theexistence of the hidden information secret [4]. Compared to Cryptography, the steganography is
practically undetectable and more secure from external attackers. Both the techniques are unique in its own way
and have their own limitations and advantages. Once the presence of hidden secret is revealed or suspected the
security of the steganography fails. In few practical applications, the strength of steganography can be enhanced
by combining it with cryptography.
The other two techniques that are similar to steganography are watermarking and fingerprinting [5].
These two techniques mainly concentrate on protecting the intellectual property and have different algorithms
and requirements compared to steganography. In watermarking the instances that are present in the image are
“marked” in a uniform manner. A signature or sign are hidden in objects in a uniform manner in order to specify
the origin or signify the ownership and protection of copyright [6]. In fingerprinting, unique marks are
embedded well defined copies of the carrier object and then they are supplied to different customers. This
enables the intellectual property owner to identify if there is any break in licensing by the customers and supply

A Design Of Novel Algorithm For Image Steganography Using Discrete Wavelet…
to other third party people [5]. In both watermarking and fingerprinting the fact that the secret message hidden
inside the object may be publicly visible and known fact [4]. But in steganography the fact that there is a secret
message hidden inside a file itself will be a secret. Depending on the type of cover object that is used in
steganography, it can be classified into text, audio, video, image and protocol steganographic techniques in
digital mediums as shown in Fig. 1.
Fig. 1. Types of Steganography in digital medium.
A. Organization of the paper
This paper is structured as follows. In section II we have given an overview of the related work that has
been done on steganography techniques. The definitions that are commonly used and the general block diagram
of Image Steganography are presented in section III. The techniques that are used are given descriptively in
section IV. A brief introduction about the algorithms and hardware/software used for the implementation is
given in section V. Finally the performance analysis is discussed and concluded in section VI and VII
respectively.
II. Related Work
Po-Yueh Chen and Hung-Ju Lin., [7] have effectively worked on a DWT based methodology for image
steganography. They have developed a new technique in which the secret message can be embedded in a
frequency domain. The new algorithm has been divided into two modes and five cases in their novel approach
for steganography. In this approach they have the secret information are embedded in the high frequency
coefficients which are resulted from the discrete wavelet transformation. Image quality is improved by
preserving the low frequency coefficients without making any modifications.
Mohammad Ali Mahrabi et al., [8] has done a remarkable work on steganalysis which are based on
statistical moments of wavelet sub band histograms. The wavelet sub bands which are derived from an image
consist of both least and most significant bits. Some of the least significant bits from grey level and some most
significant bit planes are removed and the image is decomposed using 3 levels Haar DWT. This decomposition
of image by 3 levels Haar DWT gives 13 sub bands in which the image itself is considered as LL band and
Fourier transform of each histogram sub band is calculated separately. This work gives improved detection rate
for LSB steganography compared to other techniques that are present today.
Prof U.L.Kulkarni et al., [9] has done a significant work on steganography using biometrics. Skin tone
region of images has been considered in their work to carry out the steganography using biometrics. They have
hidden the secret information in the skin tone region which is statistically undetectable compared to other
regions. The skin tone detection is performed by HSV color space. The secret information is hidden using DWT
approach which is more efficient that other frequency domain approaches. This work gives an insight into object
oriented steganography which illustrates higher security compared to other conventional methodologies.
Wang Yan and Ling-Di Ping., [10] has done an in depth study on steganography techniques and has
found a new algorithm which is based on spatial domain. They have formulated a new algorithm from which we
can hide a large amount of secret information in BMP image. They have used a methodology called as fixed
LSB’s substitute method which will compensate for distortion. This proposed method gives high capacity and
good quality of images compared to other steganography techniques that are in the present day.
K. Kanimozhi et al., [11] has performed a steganography based on dual transform using wavelets by
statistical methodology. This method extracts either DWT or IWT coefficients of both cover and secret image.
Fusion processing techniques is used on the coefficients that are extracted and stego image is obtained by the
application of various combinations of DWT and IWT on cover and secret images. Visual effect and robustness

of the stego image is corrected and performance analysis is done. This method has resulted in good
imperceptibility and robustness.
Neha Gupta et al., [12] have made a thorough study on steganography techniques and has proposed an
effective audio steganography by the use of DWT technique in this research work. This work intends to perform
the robustness and security of the audio steganography by the use of discrete wavelet transformation and LSB
techniques. The cover media they have used is an audio and can be easily used to communicate in the digital
media.
Prabakaran G and Bhavani R., [13] has done a unique digital image steganography based on the DWT
approach. They have worked on a modified approach for steganography where they have used a large size secret
image to hide in small size cover image. The secret image is scrabled by a technique called as Arnold
transformation. The DWT is performed on both cover image and secret image followed by Alpha blending
operation. Inverse DWT is applied to get the original stego image.
III. Definitions And Block Diagram
The commonly used definitions and general block diagram of Image steganography are discussed here.
B. Definitions
 Cover Image: It is an object which contains stream of data or signals which is used as a carrier of the
embedded information. The important factor of this cover image is the amount of secret information that
can be embedded into it.
 Payload Image: It is the image that is used to embed into the cover image as secret information.
 Stego Image: It is image that is obtained after the unification of cover image and payload image.
 Capacity: It is defined as the amount of secret information that can be hidden inside the cover image
without harming the properties of the cover image.
 Security: It is the measure of protection of payload image that is hidden inside the cover image.
C. General Block Diagram of Image Steganography
Fig. 2. General block diagram of Image Steganography
The general block diagram of image steganography is shown in Fig. 2. Image Steganography consists
of two individual modules namely the embedding module and the retrieval module. The embedding module is
used at the sender end where the payload image is embedded into cover image. After the successful embedding,
the stego image is derived at the end by the use of any steganographic techniques. The retrieval module is used
at the receiver end in order to extract the payload image which is embedded in the cover image by a method
called as inverse steganographic process. At the sender end encoder is used for the embedding of payload image
into cover image. Similarly, at the receiver end decoder is used to extract the payload image from stego image.

IV. DWT And Modified AES
D. Discrete Wavelet Transformation
Discrete Wavelet transform is one of the popular methods used in Image Processing domain. It has the
capacity to capture both time-frequency information of the image. The image data can be decomposed by the
method of Discrete Wavelet Transformation into different frequency ranges. This decomposition helps us to
separate frequency components that are induced by internal deformations or external factors. These decomposed
frequency components are called as sub bands. Wavelet transformation method ignores all those unwanted sub
bands and focus only on sub bands which contain maximum image information. 1 level DWT is computed by
performing successive low pass and high pass filtering of the image coefficient either in row by row or column
by column as shown in Fig. 3.
Fig. 3. Block Diagram of 1 Level DWT
Whereas in 2D DWT, the successive low pass and high pass filtering of image coefficients is done in
both row by row and column by column. This results in the decomposition of image into four sub bands namely
LL, LH,HL HH sub bands which corresponds to horizontal, vertical and diagonal features respectively. The LL
sub band contains the maximum amount of image information since it is approximately at half the original
image. The LH and HL sub bands contains the changes of images. The last sub band HH contains the detail
when the image is in high frequency. The 2D DWT and its sub bands is as shown in the Fig. 4.
Mathematically, Ψ(t) is considered as the mother wavelet and the parameters s and τ are the scaling and
shift parameters respectively. Then the 2D DWT of m x n image is given by the expression (1)
DWT (j,k) =
𝟏
𝟐𝐣
𝐟 𝐱
∞
−∞
𝛗
𝐱
𝟐
− 𝐤 𝐝𝐱 (1)
Where k is the constant of the filter and j is the power of binary.
In this research work, we have applied 1 level DWT on the cover image by using Biorthogonal DWT
as the mother wavelet. Approximation band that is LL sub band contains the major information of the cover
image. Thus, the payload image is embedded in the remaining detailed bands of the image. The wavelet
computations can be explained in three easy steps as explained below.
 Splitting: In this step, the total numbers of elements present are grouped into even and odd components that
is as shown in equation (2).
Xe = S1 = {S1, S2, S3 …….SN/2} and
XO = D1 = {d1, d2, d3……..dN/2} (2)
 Prediction: In this step, the predicted values are obtained using the equation (3).
dN/2 = dN/2 – [0.5(Sm+Sm+1) + 0.5] (3)
Where, m varies from 1 to N/2 and D1 is given by
D1 = {d1, d2, d3……..dN/2}
 Updation: In this step, the updated values are obtained by the equation (4).
SN/2 = SN/2 – [0.25(dp+dp+1) + 0.5] (4)
Where, p varies from 1 to (N/2+1) and S1 is given by
S1 = {S1, S2, S3 …….SN/2}

Fig. 4. Four Sub bands obtained after 2D DWT
E. Modified AES
The Advanced Encryption Standard (AES) with few modifications are used to encrypt the payload
image which is used as secret information and hidden into the three sub bands LH, HL and HH obtained by the
process of DWT. The modified algorithm generally consists of 11 rounds in which the first round is termed as
Add round key stage followed by nine different rounds going through six stages of operation known as iteration.
The six stages of the each iteration are as listed below.
 Sub Bytes
 Shift Rows
 Mix Columns
 Add Round Key
 Transpose of a matrix
 BIT Manipulation
The final tenth round which gives the iteration of 3 rounds is
 Sub Bytes
 Shift Rows
 Add Round Key
The only modifications that are done in modified AES algorithm is adding two more additional steps
that is transpose of a matrix and BIT manipulation. Transpose of a matrix is done after the mix row columns
where each 4*4 matrix is transposed with a intention to improve the security. Inverse transpose is applied during
the decryption of the same 4*4 matrix. In BIT manipulation the position of BITS of the transpose matrix are
interchanged. The pictorial representation of the modified AES technique is as shown in the Fig. 5. This
modification in AES algorithm improves the efficiency of the encryption and makes it more safe and secure.
The operation of each stage can be understood by the analysis of Advanced Encryption Standard by Joan
Daemen and Vincent Rijmen [14].
V. Proposed Algorithm and Resources Used
The stego image is obtained by embedding the payload image into the cover image by the
steganographic technique. We make a few assumptions like the cover and payload images are color images with
different dimensions and there is a proper channel for the transmission of stego image. The objectives of this
algorithm include improving the PSNR value and increasing security. The embedding algorithm is shown in
Fig.6 and Table I respectively and reconstruction algorithm is shown in Fig.7 and Table II. Implementation of
this image steganography is done using Beagle Board-XM with OpenCV-2.4.2 platform. Beagle board is a low
cost and low power hardware which has several facilities like 512MB RAM, USB ports, Ethernet ports and
memory card slots to store images. Implementation of steganography on Beagle Board helps it in the usage of
real time applications. Since we are using OpenCV platform to implement this, the cost of the resources are
greatly reduced and the speed of operation is improved drastically. Microsoft Visual studio Express 2012 for
Windows Desktop is used to cross check the results of PSNR and capacity values that are obtained in the
steganographic process.

Fig. 5. Modified AES Algorithm
Fig. 6. Embedding Algorithm Fig. 7. Reconstruction AES Algorithm
TABLE I. Steps for Embedding Algorithm

TABLE II. Steps for Reconstruction Algorithm
VI. Performance Analysis
The performance parameters that are considered here are PSNR value and capacity of the image. The PSNR
value of any image can be found by using the expression (5).
PSNR = 10*(log 10 ((255*255)/MSE)) (5)
Where MSE = 𝐝𝐢𝐟𝐟/(𝐡𝐞𝐢𝐠𝐡𝐭∗ 𝐰𝐢𝐝𝐭𝐡)
The cover images and the payload image that are used for the performance analysis is as shown in Fig.
8, Fig. 9, Fig. 10, Fig. 11, Fig.12 and Fig. 13. At first, we have used a constant cover image of 512 x 512
dimensions and payload image of different dimensions in order to analyze the results. The PSNR values for
different capacity using Biorthogonal DWT and Modified AES are tabulated in the Table III.
Similarly in the next stage we have used constant payload image of 256 x 256 dimensions and different
cover images of 512 x 512 dimensions to see the variations in PSNR values. It is noted that the PSNR values
slightly changes with the use of different cover images. The PSNR values are tabulated in Table IV. The PSNR
values that are obtained from the proposed algorithm are compared with other existing steganographic
techniques presented by Mohammad Reza Dastjani [15], Wang Yan [11], Kannimozhi [12], and Neha Gupta
[13]. It is observed that the PSNR value obtained in our proposed algorithm is much better compared to the
values obtained in other technologies. The comparison table of PSNR values is shown in Table V and the
screenshot obtained in Microsoft Visual studio Express 2012 during steganography process is given in Fig. 10.
TABLE III. PSNR Values for different capacity
Payload Image Size Capacity PSNR
256*256 0.2500 49.1819
400*400 0.6103 49.1704
450*450 0.7724 49.1758
470*470 0.8426 49.1623
From the above table consisting of variations in PSNR values corresponding to different capacity it can be
observed that as the capacity is increased the value of PSNR decreases.
Fig. 8. Cover Image: asheyes.jpgFig. 9. Cover Image: hfingers.jpg Fig. 10. Cover Image: dhands.jpg
(512 x 512) (512 x 512) (512 x 512)

Fig. 11. Cover Image: sfeet.jpgFig. 12. Cover Image: jears.jpg Fig. 13. Payload: navi4.jpg
(512 x 512)(512 x 512) (Different Sizes)
TABLE IV. PSNR values for different Cover Images
Cover Images Capacity PSNR value
hfingers.jpg (512 x 512) 0.250000 49.1956
dhands.jpg (512 x 512) 0.250000 49.1616
sfeet.jpg (512 x 512) 0.250000 49.1659
jears.jpg (512 x 512) 0.250000 49.1527
TABLE V. Comparison of PSNR values
Author Proposed Technique PSNR value
Mohammad Reza 2D Haar DWT 25.176
Wang Yan Spatial Domain 41.411
Kannimozhi Coefficients 40.850
Neha Gupta 1D Haar DWT 41.220
Proposed Method Biorthogonal+MAES 49.181
Fig. 10. Screenshot of steganography process in Ubuntu using OpenCV and Beagle Board-XM
VII. Conclusion
A secret data that has to be hidden and sent from the sender to receiver end is done by carefully hiding
the data inside a stego image. A secure channel is used for the transformation of this stego image from one end
to the other. The embedded data inside the stego image is highly secure because of the encryption done by the
method of modified advanced encryption technique and Biorthogonal DWT for the creation of sub bands. The
implementation of this complete steganographic process is done by using Open CV platform and Beagle Board-
XM. Applications of this image steganography include major e-commerce industries and military fields. Future
work includes usage of other complex encryption techniques and wavelet transformations for better accuracy
and security.

Acknowledgments
I express my deepest gratitude and sincere thanks to my guide Dr. Manjunath H V, Professor, Dept of
ECE, Dayananda Sagar Institutions, for his valuable time and guidance throughout this work. I am immensely
grateful to Ms. Poonam Sharma, Mr. Sathish S.B, Mr. Mahesh and Mr. Manjunath M and Dayananda Sagar
Institutions for their continuous support. I am highly indebted to my dad for his immense love and trust on me
throughout my journey of life. I also thank all my M.Tech classmates & beloved friends (JASHD) for their
continuous support and motivation throughout my career.
References
[1]. T. Moerland, “Steganography and Steganalysis,” Leiden Institute of Advanced Computing Science, www.liacs.nl/home/tmoerl/privtech.pdf
[2]. J. Silman, “Steganography and Steganalysis: An overview,” Sans Institute, 2001.
[3]. T. Jamil, “Steganography: The art of hiding information is plain sight,” IEEE Potentials, 18:01, 1999.
[4]. H. Wang and S. Wang, “Cyber Warfare: Steganography vs Steganalysis,” Communications of the acm, 47:10, October 2004.
[5]. R.J. Anderson and F.A.P. Petitcolas, “On the limits of Steganography,” IEEE Journal of selected areas in communications, May 1998.
[6]. L.M. Marvel, C.G. Jr Boncelet and C. Retter, “Spread Spectrum Steganography,” IEEE Transactions on image processing, 8:08,1999.
[7]. Po-Yueh Chen and Hung-Ju Lin, “A DWT based approach for image steganography,” International Journal of Applied Science and
Engineering, 2009.
[8]. Mohammad Ali, Mehrabi, Hassan Aghaeinia and Mojtaba Abolghasemi, “Image Steganalysis based on statistical moments of wavelet
subband histogram of images with least significant bit planes,” Congress on Image and Signal Processing, 2008.
[9]. Professor U.L. Kulkarni and Andaljali A Shejul, “A DWT based approach for Steganography using Biometrics,” International Conference
on Data Storage and Data Engineering, June 2010.
[10]. Wangyan and Ling-Di Ping, “A new Steganography algorithm based on Spatial Domain,” International Journal of Symposium on
Information Science and Engineering, Volume 3, Issue 1, pp. 408-411, January 2013.
[11]. K Kanimozhi, G. Prabakaran and Dr. R. Bhavani, “Dual Transform Based Steganography Using Wavelet Families and Statistical Methods”
International Conference on Pattern Recognition, Informatics and Mobile Engineering (PRIME), 2013.
[12]. Neha Gupta and Nidhi Sharma, “DWT and LSB based Audio Steganography,” International Conference on Reliability, Optimization and
Information Technology-ICROIT, February 2014.
[13]. Prabakaran. G and Bhavani.R, “A Modified Secure Digital Image Steganography Based on Discrete Wavelet Transform,” International
Conference on Computing, Electronics and Electrical Technologies [ICCEET] ,2012.
[14]. John Daemen and Vincent Rijmen, “The design of Rijndael: AES-The Advanced Encryption Standard,Springer 2002, ISBN: 3-54042580-2.
[15]. Mohammad Reza Dastjani and Farahani Andali Pourmohammad, “A DWT based perfect secure and high capacity Image Steganography
method,” International Conference on Parallel and Distributed Computing, Applications and Technologies, 2013.
Author Biographies
Dr. Manjunath H Vreceived his Bachelors of Engineering from Mysore University and completed his Masters
of Science (Engineering) and Doctor of Philosophy degrees from the reputed Indian
Institute of Science (IISc), Bangalore, in the field of Power Electronics and drives. He has
more than 32 years of experience in the field of Electrical Engineering and has worked in
numerous educational and research institutes globally. With his unique nature of discipline
and conduct he has contributed enormously in the field of teaching and remains as a
motivation for hundreds of students. He has worked in many topnotch research projects and
has published more than 20 innovative papers in International Conferences and has 5
globally accepted papers in reputed International Journals. Currently he is working as a
senior professor in the department of Electronics and Communication, Dayananda Sagar College of
Engineering, India. Contact: hvmanju4655@yahoo.co.in
Nagarjun B received his Bachelors of Engineering in the field of Electronics and Communication in the year
2013 from Visvesvaraya Technological University. He started pursuing his Masters of
Technology in the field of VLSI and Embedded Systems from the year 2013-2015. He has
worked on several research projects and was among the first 51 participants from India to
complete the Semiconductor Manufacturing Course from Indian Institute of Technology
(IIT-Bombay) in the year 2012. His keen interest towards technology has driven him to
participate in 4 National/International Workshops and 3 International conferences across the
country. He has presented papers in 3 National conferences and possesses 2 globally
accepted papers in International Journals. He is also a member of International Association
of Engineers &Universal Association of Computers and Electronics Engineers. Currently he is working towards
his master’s degree in VLSI and Embedded systems from Dayananda Sagar College of Engineering, India.
Contact: nagarjun9995@gmail.com

D010312230

More Related Content

What's hot (17)

Viewers also liked (19)

Similar to D010312230 (20)

More from IOSR Journals (20)

Recently uploaded (20)

D010312230