A NOVEL IMAGE STEGANOGRAPHY APPROACH USING MULTI-LAYERS DCT FEATURES BASED ON SUPPORT VECTOR MACHINE CLASSIFIER

The International Journal of Multimedia & Its Applications (IJMA) Vol.9, No.1, February 2017
DOI : 10.5121/ijma.2017.9101 1
A NOVEL IMAGE STEGANOGRAPHY APPROACH
USING MULTI-LAYERS DCT FEATURES BASED ON
SUPPORT VECTOR MACHINE CLASSIFIER
Akram AbdelQader1
and Fadel AlTamimi2
1
Department of Multimedia Systems, AL-Zaytoonah University Of Jordan, Amman,
Jordan
2
Department of Computer Science, AL-Zaytoonah University Of Jordan, Amman,
Jordan
ABSTRACT
Steganography is the science of hidden data in the cover image without any updating of the cover image.
The recent research of the steganography is significantly used to hide large amount of information within
an image and/or audio files. This paper proposed a new novel approach for hiding the data of secret image
using Discrete Cosine Transform (DCT) features based on linear Support Vector Machine (SVM)
classifier. The DCT features are used to decrease the image redundant information. Moreover, DCT is
used to embed the secrete message based on the least significant bits of the RGB. Each bit in the cover
image is changed only to the extent that is not seen by the eyes of human. The SVM used as a classifier to
speed up the hiding process via the DCT features. The proposed method is implemented and the results
show significant improvements. In addition, the performance analysis is calculated based on the
parameters MSE, PSNR, NC, processing time, capacity, and robustness.
KEYWORDS
Discrete Cosine Transform (DCT), Support Vector Machine (SVM), Classifier, Steganography, Peak
signal-to-noise ratio (PSNR).
1. INTRODUCTION
Recently, the growth of technology and social media communications over the internet motivate
the researchers to develop new steganography techniques. The objective of steganography is to
hide a secret message within a cover-media in such a way that others cannot discern the presence
of the hidden message. Today, steganography technology and steganalysis are attracted on much
attention. Steganalysis is focusing on the detection the present of hidden message. In the
literatures, many steganalytic research and schemes for digital images have been proposed [1]–
[3]. Recently, the security issues are very important research due to the wide amount of
information over the internet.
Steganography in the simple means hidden date in other, such as image, audio file or even a video
file [2]. An image is one type of steganographic, where the secret image is hidden in a cover
image based on some hiding algorithm. Form the state of the arte, many researchers use Least
Significant Bits (LSB) to reduce the distortion of the stegano image [22]. The original image used

2
to hide image is called a cover image in steganography, and the image to hide is called a secret
image [3].
The objective of steganography is hiding the data into the cover image such that the existence of
data in the cover image is not seen to the human beings [4]. The figure 1.1 shows the process of
hiding data.
Figure1. The process of hiding data
Many techniques were proposed to implement steganography such as Spatial-Domain methods
(LSB) where processing is applied directly on the pixel values of the image [22] and [23],
Transform Domain Methods (DCT) and Discrete Wavelet Transform (DWT) technique pixel
values are transformed and then processing is applied on the transformed coefficients, and
Statistical Methods [24] (Syndrome Trellis codes).
2. DISCRETE COSINE TRANSFORM (DCT)
Discrete Cosine Transform (DCT) techniques are used in frequency domain. The DCT is a
function that convert data from the spatial domain to the frequency domain. In DCT, after
converting the image in frequency domain, the data is hidden in the least significant bits of the
medium frequency components, secret messages are hidden in the high frequency coefficients
resulted from Discrete Wavelet Transform and provide maximum robustness. In the Least
Significant Bit (LSB), every pixel of an image is converting into the (1) or (0) and data is hidden
into the least significant position of the binary value of the pixels of the image.
DCTs are important to numerous applications in science and engineering, from lossy compression
of audio and images (where small high-frequency components can be discarded), to spectral
methods for the numerical solution of partial differential equations. The use of cosine functions is
critical in these applications: for compression, it turns out that cosine functions are much more
efficient (as described below, fewer functions are needed to approximate a typical signal),
whereas for differential equations the cosines express a particular choice of boundary conditions.
The most common variant of discrete cosine transform is the simply DCT, in addition to the
modified discrete cosine transforms (MDCT), which is based on a DCT of overlapping data.
[6][7]. Multidimensional variants of the various DCT types follow straightforwardly from the
one-dimensional definitions: they are simply a separable product (equivalently, a composition) of
DCTs along each dimension. Figure 2 shows a Discrete Cosine Transform of an Image.

3
Figure 2 Discrete Cosines Transform of an Image.
Where u= 0, 1, 2….. N-1
The general equation for a 2D (N by M image) DCT is defined by the following equation:
where u, v = 0,1,2….N-1
Here, the input image is of size N X M. c (i, j) is the value of the pixel in row i and column
j; C(u,v) is the DCT coefficient in row u and column v of the DCT matrix. DCT is used into
steganography as [8]. Image is segment into 8×8 blocks or 4X4 block of pixels. Working
from left to right, top to bottom, DCT is applied for each block. Each block is compressed
through quantization table to scale the DCT coefficients and message is hidden in DCT
coefficients. Figure 3 shows a two dimensional DCT frequencies from the RGB DCT image
of 8 X8 pixels.
Figure 3 Two-dimensional DCT frequencies from the RGB DCT image based on 8 X8 pixels
3. SUPPORT VECTOR MACHINE
Support vector machines (SVM) also support vector networks [25] are supervised learning
models with associated learning algorithms that analyses data used for classification and
regression analysis. Based on training example set, each marked as belonging to one of two

4
categories, since SVM builds a model that assigns new examples to one category or to the other.
In Linear SVM a training dataset of n points of the form as in the figure 4 below
where the yi are either 1 or −1, each indicating the class to which the point belongs. Each
is a p-dimensional real vector. We want to find the "maximum-margin hyperplane that divides the
group of points for which from the group of points for which , which is
defined so that the distance between the hyper-plane and the nearest point from either group
is maximized. Any hyperplane can be written as the set of points satisfying
where is the (not necessarily normalized) normal vector to the hyperplane. The
parameter determines the offset of the hyperplane from the origin along the normal vector .
Figure 4 Linear SVM classifier
4. THE LITERATURE REVIEW
Many researches were done on Steganography because it is very important in computer
multimedia fields, internet fields and in security systems. In addition, it is very important to know
how much data can be concealed without image distortion. In literature, the techniques of
steganography and its implementation were explained by J.R.Krenn in [1].
In [5], Chen Ming, et. al. focused on the steganography tools algorithms. Based on the analyses of
the algorithms, various tools are divided into five categories: (1). Spatial domain based
steganography tools; (2). Transform domain based steganography tools; (3). Document based
steganography tools; (4) File structure based Steganography tools; (5) other categories, e.g. video
compress encoding and spread spectrum technique based. Deshpande Neeta, et. al. [9], they
proposed the Least Significant Bit embedding technique suggests that data can be hidden in the
least significant bits (LSB) of the cover image and the human eye would be unable to see the
hidden data in the cover image. They explained the LSB embedding technique and presents the
evaluation results for 2, 4, 6 (LSB) for a PNG images and a .bmp images.
In [10] K.B.Raja, et. al., they proposed a challenging task of transferring the hidden data to the
cover without being detected. In addition, they used compression techniques on raw images to

enhance the security of the payload. Vijay Kumar Sharma,
steganography algorithm based on
operation to ensure the security against the steganalysis attack.
proposed a new steganography technique
and image quality, they embeds the secret messages in frequency domain
was divided into two modes and 5 cases. Ane
bitmap images with almost no perceptible difference between the original image and
image. Others in [14] they discusse
proposed a novel technique to hide data in a colorful image using least significant bit. Hassan
Mathkour,et. al. in [15] they
presented steganography techniques
that takes the advantages of the strengths and avoids the limitations
Finally, in [17] MamtaJuneja, et. al.
based on LSB and the RSA encryption t
5. THE PROPOSED APPROACH
In this paper we use the SVM classifier for training DCT Appling the linear SVM algorithm.
DCT2 is used as shown in the equation 1
Equation (1) DCT2 algorithm.
5.1 DCT FEATURES EXTRACTION
AND COVER IMAGES ALGORITHM
The proposed system has been designed to hide the secret message in the cover image
following algorithm shows the process of hidden a message:
The proposed algorithm 1 and methodology is shown in figure 5 and
following steps:
enhance the security of the payload. Vijay Kumar Sharma, et. al. [11] the proposed a
based on 8bit gray scale or 24bit color image, they used the
operation to ensure the security against the steganalysis attack. Other researchers in
proposed a new steganography technique based on different users demands on hiding capacity
embeds the secret messages in frequency domain the proposed algorithm
divided into two modes and 5 cases. Aneesh Jain, et. al. [13] they were hide the data in a
almost no perceptible difference between the original image and
discussed a survey of general steganography techniques and they
echnique to hide data in a colorful image using least significant bit. Hassan
in [15] they analyzed and evaluated the strengths and weaknesses of the
presented steganography techniques. In addition, they proposed a robust steganography technique
advantages of the strengths and avoids the limitations of the studied techniques
MamtaJuneja, et. al. they proposed a Robust image steganography technique
RSA encryption technique.
PPROACH
DCT2 is used as shown in the equation 1.
XTRACTION AND SVM TRAINING PROCESS FOR THE
LGORITHM.
The proposed system has been designed to hide the secret message in the cover image
shows the process of hidden a message:
and methodology is shown in figure 5 and in our algorithm we use the
5
the proposed a new
they used the logical
r researchers in [12] they
based on different users demands on hiding capacity
the proposed algorithm
they were hide the data in a
almost no perceptible difference between the original image and the result
techniques and they
echnique to hide data in a colorful image using least significant bit. Hassan
the strengths and weaknesses of the
robust steganography technique
of the studied techniques.
a Robust image steganography technique
HE ORIGINAL
The proposed system has been designed to hide the secret message in the cover image, the
n our algorithm we use the

6
Figure 5 the proposed algorithm and methodology
The proposed algorithm and methodology are tested and the results are shown the following
figures. Figure 6 shows the cover original image with size 256 *256 which used to hide the
message. Figure 7 shows the original message that used to hide in the cover image of size of 64*
64. Figure 8 shows the stego image after hiding the secrete in the cover image.

7
Figure 8 Stego image size 256*256
The proposed algorithm is implemented and tested and shows a significant results. To show the
algorithm accuracy and performance many comparisons is done. The PSNR comparison is PSNR
44.3058 dB and the histogram between the original and stego images as shown in figure 9 and
figure 10.
Figure 10 Histogram of cover image and stego image

8
5.2 IDCT FEATURES EXTRACTION AND SVM TRAINING PROCESS FOR THE STEGO
IMAGE ALGORITHM.
The next step of the proposed algorithm is to extracting the secret image from the stego image as
shown in Algorithm 2 and in the methodology in figure 11.
Figure 10 the proposed algorithm to regenerate the secret image from the stego image
6. SYSTEM IMPLEMENTATION
This system was implemented using Matlab programming language over a PC Core i7 CPU with
8 GB RAM on Windows 7 Operating System. Many other applications were running such as
anti-virus application which may be effect the performance rate. The proposed system is tested
for many images width different sizes and different resolutions, and it shows significant results
based on DCT and SVM classifier. The proposed algorithm was tested for processes; the process
of hiding the secret message in the cover image and the process of retrieving the secret image
from the stego image and show good results.
The proposed system is implemented and tested using many images width different sizes and
different resolutions, and it shows significant results using the proposed three layer DCT methods
and SVM classifier.

9
Many experiments were done using the proposed algorithm and the results are analyzed
and reported. Table 1 shows the comparisons of PSNR between cover image and stego
image. In addition table 2 shows the comparisons of PSNR between secret and message
image.
Table 1 Result of PSNR between cover and stego image
Table 2 Result of PSNR between secret and message image
The tables 1 and 2 show a significant improvement results based on the ratio between the
maximum possible power of a signal and the power of corrupting noise that affects the fidelity of
the two compared images.
7. CONCLUSIONS
The proposed approach shows significant results based DCT features and the using of fast linear
SVM classifier that used in both processes (hiding and retrieving). Moreover, the using of the
three layer of a color image based on RGB over DCT features add significant improves in the
performance and the accuracy. The future work may be done by using other classifier and other
features and compare the results.
REFERENCES
[1] J.R. Krenn, “Steganography and Steganalysis”, January 2004.
[2] BeenishMehboob and Rashid Aziz Faruqui, “A Steganography Implementation”, IEEE -4244-2427-
6/08/ 2008.
[3] M. M. Amin, M. Salleh, S. Ibrahim, M.R.Katmin, M.Z.I. Shamsuddin, “Information Hiding using
Steganography” Proceedings of 4th National Conference on Telecommunication Technology, Shah
Alam, Malaysia, 2003.
[4] K Suresh Babu, K B Raja, Kiran Kumar K, Manjula Devi T H, Venugopal K R, L M Patnaik,
“Authentication of Secret Information in Image Steganography”.
[5] Chen Ming, Zhang Ru, NiuXinxin, Yang Yixian, “Analysis of Current Steganography Tools:
Classifications & Features” , International Conference on Intelligent Information Hiding and
Multimedia Signal Processing (IIH-MSP'06),IEEE- 0-7695-2745-0/06 2006.
[6] Ahmed, N.; Natarajan, T.; Rao, K. R. (January 1974), "Discrete Cosine Transform", IEEE
Transactions on Computers C–23 (1): 90–93, doi:10.1109/T-C.1974.223784.
[7] Rao, K; Yip, P (1990), Discrete Cosine Transform: Algorithms, Advantages, Applications, Boston:
Academic Press, ISBN 0-12-580203-X
[8] NageswaraRaoThota, Srinivasa Kumar Devireddy, “Image Compression Using Discrete Cosine
Transform”, Georgian Electronic Scientific Journal: Computer Science and Telecommunications,
No.3 (17), 2008.

10
[9] Deshpande Neeta, KamalapurSnehal, Daisy Jacobs, “Implementation of LSB Steganography and Its
Evaluation for Various Bits”, 2004.
[10] K.B.Raja, C.R.Chowdary, Venugopal K R, L.M.Patnaik, “A Secure Image Steganography using LSB,
DCT and Compression Techniques on Raw Images”, IEEE-0-7803-9588-3/05 2005.
[11] Vijay KumarSharma, Vishalshrivastava, “A Steganography Algorithm for Hiding Images by
improved LSB substitution by minize detection.”Journal of Theoretical and Applied Information
Technology, Vol. 36 No.1, ISSN: 1992-8645, 15th February 2012.
[12] Po-Yueh Chen and Hung-Ju Lin, “A DWT Based Approach for Image Steganography”,International
Journal of Applied Science and Engineering 4, 3: 275-290, 2006.
[13] AneeshJain,IndranilSen. Gupta, “A JPEG Compression Resistant Steganography Scheme for Raster
Graphics Images”,IEEE-1-4244-1272-2/07 2007.
[14] BeenishMehboob and Rashid Aziz Faruqui, “A Steganography Implementation”, IEEE -4244-2427-
6/08 2008.
[15] Hassan Mathkour, Batool Al-Sadoon, AmeurTouir, “A New Image Steganography Technique”,IEEE-
978-1-4244-2108-4/08 2008.
[16] NageswaraRaoThota, Srinivasa Kumar Devireddy, “Image Compression Using Discrete Cosine
Transform”, Georgian Electronic Scientific Journal: Computer Science and Telecommunications,
No.3 (17), 2008.
[17] MamtaJuneja,Parvinder Singh Sandhu, “Designing of Robust Image Steganography Technique Based
on LSB Insertion and Encryption”, International Conference on Advances in Recent Technologies in
Communication and Computing, 2009.
[18] Vijay Kumar, Dinesh Kumar, “Performance Evaluation of DWT Based Image Steganography”
,IEEE- 978-1-4244-4791-6/10 2010.
[19] Ali Al-Ataby and Fawzi Al-Naima, “A Modified High Capacity Image Steganography Technique
Based on Wavelet Transform” The International Arab Journal of Information Technology, Vol. 7, No.
4, October 2010.
[20] T. Narasimmalou, Allen Joseph .R, “Optimized Discrete Wavelet Transform based Steganography” ,
IEEE International Conference on Advanced Communication Control and Computing Technologies
(ICACCCT),2012.
[21] NedaRaftari and Amir MasoudEftekhariMoghadam, “Digital Image Steganography Based on
Assignment Algorithm and Combination of DCT-IWT”, Fourth International Conference on
Computational Intelligence, Communication Systems and Networks, 2012.
[22] S. M. Mohadeseh, N. Hossein, "The pair-wise LSB matching steganography with a discrete quantum
behaved Gravitational Search Algorithm", Journal of Intelligent & Fuzzy Systems, vol. 30, no. 3, pp.
1547-1556, 2016.
[23] Mamta Juneja and Parvinder Singh Sandhu, “A New Approach for Information security using an
Improved Steganography Technique”, Journal of Info.Pro.Systems, Vol 9, No:3, pp.405-424, (2013) .
[24] Tomas Filler, Student Member, IEEE, Jan Judas and Jessica Fridrich, Member, IEEE, (2010)
“Minimizing Additive Distortion in Steganography using Syndrome Trellis Codes”, IEEE Article,
pp.1-17.
[25] Cortes, C.; Vapnik, V. "Support-vector networks". Machine Learning. 20 (3): 273–297.
doi:10.1007/BF00994018, (1995).

A NOVEL IMAGE STEGANOGRAPHY APPROACH USING MULTI-LAYERS DCT FEATURES BASED ON SUPPORT VECTOR MACHINE CLASSIFIER

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to A NOVEL IMAGE STEGANOGRAPHY APPROACH USING MULTI-LAYERS DCT FEATURES BASED ON SUPPORT VECTOR MACHINE CLASSIFIER (20)

Recently uploaded (20)

A NOVEL IMAGE STEGANOGRAPHY APPROACH USING MULTI-LAYERS DCT FEATURES BASED ON SUPPORT VECTOR MACHINE CLASSIFIER