A Mask Based Approach For Lossless Colour Image Steganography.

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

A Mask based approach for Lossless Color Image Steganography

A Mask based approach for Lossless Color


Image Steganography
Emergent Research Forum

MaheshKumar H. Kolekar Abhinash Kumar Jha


Department of Electrical Engineering The LNM Institute of Information
Indian Institute of Technology, Patna Technology, Jaipur.
India. India.
email : mahesh@iitp.ac.in email : abhijha.lnm@gmail.com

Apurva Singh Sneha Srivastava


The LNM Institute of Information The LNM Institute of Information
Technology, Jaipur. Technology, Jaipur.
India. India.
email : anjusrivastava68@gmail.com email : asinghvirgo@gmail.com

Abstract
Steganography is the art or practice of concealing a message, image, or file within another message,
image, or file. The main advantage of Steganography over cryptography is that an ordinary observer
cannot tell whether the file contains any hidden information in it or not. A form of Steganography where
the original message is retrieved without any loss in data is called Lossless Image Steganography. In this
research paper we deal with hiding a digital message image inside a digital cover image leading us to the
stego image. The embedding and extraction causes no loss of message making it lossless. Our method is a
form of message secure transmission since it cannot be retrieved by an attacker who is unaware of the
algorithm. The proposed algorithm has been compared with four other algorithms in terms of Peak Signal
to Noise Ratio (PSNR) showing its supremacy.

Keywords

Steganography, Lossless, Data Transmission, Digital Security, Mask based, Arnold Transform

Introduction
The Rise of internet has a huge impact over the communication. Over millions of messages are exchanged
each day in forms of images, videos, sounds, text mails, etc., leading to an increased demand of
information security. Cryptography and Steganography are two techniques used for this purpose.
Cryptography deals with the construction and analysis of protocols to overcome the presence of
adversaries [1] in various terms of information security such as confidentiality, data integrity,
authentication and non-repudiation [2]. Its major drawback is it’s inefficiency in maintaining secrecy of
the existence of message. Steganography on other hand is the technique of hidden communication
focusing on hiding the existence of a message itself. Application of Steganography is not limited to secure
data communication but ranges from copyright controls, enhancement in the robustness image search
engine and also in document forensics [3]. Steganography can also be applied for video indexing [4]-[5]
algorithms for critical data.

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 1


A Mask based approach for Lossless Color Image Steganography

In Image Steganography, the cover (or carrier) used is an image file. The embedding algorithm is used to
hide secret messages inside a cover image to make Stego-Image and is usually protected by a keyword for
prevention of access from unwanted third parties. Secret message are obtained back by the extraction
process applied at the message receiver’s end. The main goal is to not raise suspicion and avoid
introducing statistically detectable modifications into the cover image. The embedded information is
undetectable if the image with the embedded message is consistent with the model of the source from
which the carrier images are drawn. We point out that the ability to detect the presence does not
automatically imply the ability to read the hidden message. We further note that un-detectability should
not be mistaken for invisibility, a concept tied to human perception. One of the very basic methods used is
as follows: for equal sized message and cover image, in each overlapping pixel, the MSB of the message
image in placed in place of the LSB of the cover image. This ensures high PSNR of cover image. However,
this single stepped encryption is fairly easy to crack and hence never used in practice.
Upon surveying the literature several techniques can be found. Chan et al. [6] proposed the popular
Stenographic method based on the least-significant-bits (LSBs) substitution. Discrete Wavelet Transform
(DWT) used by Chen et al. [7] and Discrete Fourier transform (DFT) used by Wang et al. [8] for hiding
image in phase achieves an advantage in capacity, but vision effect is unsatisfactory. Yang [9] proposed
module substitutions based stenography method for color images embedding a variety of secret bits into,
the R, G, and B plane is encoded by Mod u, Mod u-v and Mod u-v-w substitution, respectively, to alleviate
further color distortion and obtain a larger hiding capacity leading improvement in visual quality. To
improve embedding capacity Chang et al. [10] used the concept of modulo operation and side-match-
vector-quantization (SMVQ) to conceal secret data into a pallet-based. Ghasal et al [11] proposed a
payload based steganography technique which uses DWT for embedding .Taking advantage of the human
visual system Chou et al [12] applied B plane in RGB color system and U plane in the YUV color system,
which to conceal as many secret data. Experimental results showed that they successfully achieved the
goals of the high embedding capacity and maintaining the visual quality. Hidden Markov Model (HMM)
[13]-[14] is a popular concept for statistical modelling. Sallee [15] used the HMM concept for
steganography and steganalysis, it provides us explanation for the questions as such as how large a
message can be hidden without risking detection by certain statistical methods, and how to achieve this
maximum capacity.
In many practical situations, it becomes very important that the information sent is received without any
minor (even minute) changes. Hence the necessity for lossless stenographic technique is required. Marvel
et al. [16] used a lossless image compression followed by an entropy coder for the purpose, however due to
limitations of encoders by Shannon’s theorem and classifiers but it fails at high embedding rate. A
stenographic system is insecure if an attacker is able to prove the existence of a secret message. In this
paper we implement a private key [17] based multilevel cryptography based encryption of message image
pixels in their binary form. Embedding has been performed by replacing the two LSBs of cover image by
message image pixels in a pattern designed to ensure randomness.
The organization of the rest of the paper is as follows section 2 describes the methodology while section 3
consists of the simulation results. Conclusion is drawn in section 4.

Methodology
Any stenographic technique constitutes of an initial preparation followed by an embedding procedure. An
extraction procedure is essential to extract the message from the stego image. Generalized framework of
any stenographic technique is shown in Figure 1.
Steganography technique described in the paper consists of two major steps. The first step deals with
preprocessing of cover and message image enhancing the security aspect of the algorithm. In the second
stage a mask based approach is applied for the embedding process resulting in generation of the stego
image.

A. Initial Preparation and Analysis


One of the major advantages of Steganography is freedom in the selection of cover image by embedder
that results in the least detectable stego image. Also introduction to encryption to message image before

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 2


A Mask based approach for Lossless Color Image Steganography

embedding will result in appearance of message as some noise like pattern. Taking advantage from the
above mentioned facts the step constitutes of two phases, former one on focuses on selection and
enhancement of cover image while the latter one concentrates on the security aspect of the message.

Figure 1: Structure and Working of a Steganography Algorithm

1) Selection and Analysis of Cover Image

Selection test: Using appropriate features in selection of cover image one can ensure less detectability
and enhanced security. JPEG format of image having equally distributed histogram and containing
Unrecognizable patterns are preferred for cover images due to its unresponsiveness to visual attacks.
(Visual attacks mean that Stenographic messages can be seen on the low bit planes of an image because
they overwrite visual structures; this usually happens in BMP images).
Analysis: By proper analysis of cover image using basic image processing techniques such as brightness
correction, contrast correction, tona correction and color balance taking the nature of message image into
account one can further reduce attention from the unwanted adversaries.

2) Preparation of Message Image

Arnold Transform: Arnold Transform [18] also known as Arnold’s cat map is a revertible chaotic map
from the torus into itself. One of the map’s important feature is it returns its initial state within some
finite steps if after some being apparently randomized by the transformation. As can be seen in the Figure
2, the Original image of the cat is sheared and then wrapped around in the first iteration of the
transformation. After some iteration, the resulting image appears random and disordered ultimately
returning to the original image. For an image, the transformation can be implemented by mapping given
as:

, →  
, 
 (1)

Where, I[i;j] denotes the current spatial location of the pixel in an image.

Message Preparation: To enhance the security aspect of the message an encryption mechanism which
can be called as a child function of a standard of cryptography DES [19] is use. It includes logical XOR
operation as a step function manner, which secures the message from brute force attack or direct use of
multiple logical XOR operation with the key as a guess.

B. Embedding
The embedding procedure uses a mask based approach. Mask is variable fixed sized running window that
scans and modifies image as per the requirement. Depending upon the application and requirements two
masks Mask-9 and Mask-4 are introduced in the paper.Mask-9 uses a 3 X 3 window and uses the LSB of
the cover image pixels for embedding message leading to a high quality stego image. It provides us an
minimum payload capacity of 12.5 percent in case of lossless recovery. The capacity can be further
increased by using image compression techniques.

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 3


A Mask based approach for Lossless Color Image Steganography

(a) (b)
Figure 2: (a) Linear mapping done by Arnold transform (b) Effect of Arnold transform on a
sample image

In mask-4, a 2 X 2 window is selected and the first two least significant bits of the cover image pixels are
used for embedding message giving us a minimum payload capacity of 25 percent with lossless message
recovery. Upon comparing the usefulness of both the masks the mask-4 gives us a better lossless payload
while mask-9 leads to an optimum quality stego image. Also for the same size of cover and message if both
masks are used, a higher quality will depend on the data chosen as the second mask will bring change to
less cover image pixels than the first one.

Figure 3: Flow Chart describing the Encoding Process

Flow chart as shown in Figure 3 describes the complete procedure. The image is converted into 8-bits
planes and depending on the selected mask, XOR is done. For mask-9 the each binary bit of this message
image is placed into one of the neighboring 8 pixels binary LSB while in case of mask- 4 two successive
binary bits of this message pixel are placed in the two LSBs of the neighboring pixels respectively.

C. Extraction and Recovery of Message


Recovery of message from stego image involves two major steps. First step would be to extract the
message image from the stego one and second step includes restoration message image using decryption
and inverse Arnold transform. A general overview of the whole procedure can be described as in Figure 4.

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 4


A Mask based approach for Lossless Color Image Steganography

Figure 4: Flow Chart describing the Embedding and Extraction Procedure

EXPERIMENTAL ANALYSIS

Figure 5: Data Set of Cover Images Used For Simulation

In order to demonstrate the performance of the proposed algorithm, we tested it subjectively as well as
objectively. Our dataset consists of a set of ten cover images as shown in Fig 5 and message images, each
of different sizes for comparing efficiency of our algorithm with change in size.

175X175 200X200 225X225 250X250


S. No.
1 47.22 46.14 45.00 44.24
2 47.23 46.14 45.01 44.25
3 47.24 46.12 45.03 44.23
4 47.08 46.11 45.01 44.24
5 47.12 46.11 45.02 44.02
6 47.20 51.72 45.01 51.71
7 47.22 51.72 45.01 51.72
8 47.24 51.75 45.02 51.72
9 47.26 46.15 45.02 44.26
10 47.19 46.12 45.01 44.23
TABLE 1: PSNR (In Db) Variation for MASK-4 (M4) for different message size

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 5


A Mask based approach for Lossless Color Image Steganography

Subjective measurement was carried out using Peak Signal to Noise Ratio (PSNR). PSNR is the ratio
between maximum possible power of a signal and the power of corrupting noise that affects the fidelity of
its representation and is given by the equation:

, 
= 10 (2)
 !",#

Where, MSE (O,I) is the mean square error between the output image(o) and the source image(I).
Table 1 and Table 2 give us a detailed subjective analysis in terms of PSNR for different cover images
when embedded with different size of messages.

Image 25X25 50X50 75X75 100X100 125X125 150X150


1 68.10 62.11 58.64 56.13 54.18 52.58
2 68.12 62.15 58.62 56.12 54.19 52.60
3 68.18 62.13 58.62 56.12 54.20 52.60
4 68.00 62.11 58.64 56.16 54.15 52.56
5 68.08 62.08 58.65 56.17 54.15 52.56
6 68.18 62.15 58.63 56.13 54.19 52.59
7 68.13 62.14 58.61 56.12 54.19 52.61
8 68.17 62.15 58.62 56.13 54.18 52.56
9 68.16 62.15 58.61 56.13 54.18 52.59
10 68.13 62.11 58.61 56.12 54.16 52.60
TABLE 2: PSNR (In Db) variation for MASK-9 (M9) for different message sizes

In order to put our algorithm into context we implemented four other algorithms namely highly
undetectable stego systems (HUGO)[20], wavelet obtained weights (WOW)[21], Edge Adaptive Image
Steganography Based on LSB Matching(EAISBLM)[22] and spatial universal wavelet relative distortion
(S-Uniward) [23]. The Comparative PSNR result is described in Table 3.

Image HUGO WOW EAISBLM S-UNIWARD M9 M4


1 51.70 53.80 52.36 54.45 62.11 58.24
2 51.70 53.89 52.43 54.43 62.15 58.20
3 51.71 54.42 52.22 54.59 62.13 58.27
4 51.73 52.77 52.01 54.35 62.11 57.94
5 51.75 53.19 51.85 54.15 62.09 58.27
6 51.69 53.80 52.42 54.55 62.16 58.25
7 51.46 54.19 52.14 54.57 62.15 58.22
8 51.69 54.24 51.78 54.54 62.15 58.19
9 51.77 54.14 52.43 54.52 62.15 58.33
10 50.92 53.93 52.42 54.49 62.11 58.23
TABLE 3: PSNR Comparison between different techniques for a sample cover and message
image

A visual demonstration of the resulting stego images for 4 different resolutions each of mask-9 and mask-
4 can been seen in Figure 6.

Conclusion
Use of digital images are becoming are more common in steganographic communication due to easy
availability and less suspicion. Minimizing detectability and maximizing payload is the primary target of

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 6


A Mask based approach for Lossless Color Image Steganography

steganographer now-days. Using the transform domain for the purpose would definitely provide us huge
payload but due to their poor visual quality they are detectable. In this paper we presented a mask based
approach for Steganography. Two masks, mask-4 and mask-9 are presented where mask-4 provides us a
better payload while mask mask-9 gives us an optimum visual quality. Use of Arnold transform and
Cryptographic algorithm ensures us regarding security aspect of the proposed scheme. In the
experimental simulations carried out raw color images were directly embedded using the scheme after
preprocessing. The payload capacity can be increased if we use lossless compression technique before
applying the proposed scheme. The present work can be extended in future by introducing
new masks compressing the message image simultaneously with the embedding process leading to high
payload as well as computational efficiency

Figure 6: Visual Demonstration of Stego Images.

REFERENCES
Katz, J., and Lindell, Y. (2007). Introduction to modern cryptography: principles and protocols. CRC
Press.
Menezes, A. J., Van Oorschot, P. C., and Vanstone, S. A. (1996). Handbook of applied cryptography. CRC
press.

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 7


A Mask based approach for Lossless Color Image Steganography

Cheddad, A., Condell, J., Curran, K., and Mc Kevitt, P. (2009). A secure and improved self-embedding
algorithm to combat digital document forgery. Signal Processing, 89(12), 2324-2332.
Kolekar, Maheshkumar H. "Bayesian belief network based broadcast sports video indexing." Multimedia
Tools and Applications 54.1 (2011): 27-54.
Kolekar, M. H., Palaniappan, K., Sengupta, S. “Video Event Detection and Semantic Identification Using
Bayesian Belief Networks”, IEEE Int. Conf. on Computer Vision (ICCV) workshop, Kyoto, Japan, Sept
2009
Chan, C. K., and Cheng, L. M. (2004). Hiding data in images by simple LSB substitution. pattern
recognition, 37(3), 469-474.
Chen, P. Y., and Lin, H. J. (2006). A DWT based approach for image steganography. International
Journal of Applied Science and Engineering, 4(3), 275-290.
Wang, Y., and Moulin, P. (2006, February). Statistical modelling and steganalysis of DFT-based image
steganography. In Electronic Imaging 2006 (pp. 607202-607202). International Society for Optics
and Photonics.
Yang, C. Y. (2007, November). Color image steganography based on module substitutions. In Intelligent
Information Hiding and Multimedia Signal Processing, 2007. IIHMSP 2007. Third International
Conference on (Vol. 2, pp. 118-121). IEEE.
Chang, C. C., Wu, W. C., and Chen, Y. H. (2008). Joint coding and embedding techniques for multimedia
images. Information Sciences, 178(18), 3543-3556.
Ghasal, P. S., Trivedi, P., Chandwani, A., Tetrawal, M. L., Jha, A., & Kumar, A. (2014, February). Payload
minion stenographic technique for color images. In Optimization, Reliabilty, and Information
Technology (ICROIT), 2014 International Conference on (pp. 306-309). IEEE.
Chou, Y. C., Chang, C. C., and Li, K. M. (2008). A large payload data embedding technique for color
images. Fundamenta Informaticae, 88(1), 47-61.
Kolekar, M. H., & Sengupta, S. (2005, July). Hierarchical structure for audio-video based semantic
classification of sports video sequences. In Visual Communications and Image Processing 2005 (pp.
596017-596017). International Society for Optics and Photonics.
Kolekar, M. H., & Palaniappan, K. (2008, November). A hierarchical framework for semantic scene
classification in soccer sports video. In TENCON 2008-2008 IEEE Region 10 Conference (pp. 1-6).
IEEE.
Sallee, P. (2005). Model-based methods for steganography and steganalysis. International Journal of
Image and graphics, 5(01), 167-189.
Marvel, L. M., Boncelet Jr, C. G., and Retter, C. T. (1999). Spread spectrum image steganography. Image
Processing, IEEE Transactions on, 8(8), 1075-1083.
Rabin, M. O. (1979). Digitalized signatures and public-key functions as intractable as factorization
Qi, D., Zou, J., & Han, X. (2000). A new class of scrambling transformation and its application in the
image information covering. Science in China Series E: Technological Sciences, 43(3), 304-312.
Coppersmith, D. (1994). The Data Encryption Standard (DES) and its strength against attacks. IBM
journal of research and development, 38(3), 243-250.
Pevn, T., Filler, T., and Bas, P. (2010, January). Using high-dimensional image models to perform highly
undetectable steganography. In Information hiding (pp. 161-177). Springer Berlin Heidelberg.
Holub, V., and Fridrich, J. J. (2012, December). Designing steganographic distortion using directional
filters. In WIFS (pp. 234-239).
Luo, W., Huang, F., and Huang, J. (2010). Edge adaptive image steganography based on LSB matching
revisited. Information Forensics and Security, IEEE Transactions on, 5(2), 201-214.
Holub, V., Fridrich, J., and Denemark, T. (2014). Universal distortionfunction for steganography in an
arbitrary domain. EURASIP Journal on Information Security, 2014(1), 1-1

Twenty-first Americas Conference on Information Systems, Puerto Rico, 2015 8

You might also like