2012 - International Conference on Emerging Trends in Science, Engineering and Technology 192

Data Hiding in Intermediate Significant Bit Planes, A High Capacity Blind Steganographic Technique
Shabir A. Parah
Post Graduate Department of Electronics and Instrumentation Technology, University of Kashmir Srinagar, India shabireltr@gmail.com

Javaid A. Sheikh
Post Graduate Department of Electronics and Instrumentation Technology, University of Kashmir Srinagar, India sjavaid_29ku@yahoo.co.in

G.M. Bhat
University Science Instrumentation Center, University of Kashmir, Srinagar India gmbhat_ku@yahoo.co.in

AbstractThe availability of relatively inexpensive digital products coupled with the promise of higher bandwidth and quality of service (QoS) for both wired and wireless communication networks have made it possible to create, replicate, transmit, and distribute digital data without any loss in quality. In such a scenario steganography has received huge attention from the research community round the globe, as it has been found useful for information security and under cover communication. Stegnography refers to covert communication for transfer of confidential information over a communication channel. This paper presents a high capacity stegnographic technique in which secret data is embedded in Intermediate Significant Bit planes of the cover image. The data to be embedded is broken down in blocks of relatively decreasing lengths and each block is embedded in the cover media under control of a highly secure key. This work shows attractive results with respect to imperceptibility and capacity when compared with a few reported techniques in addition to providing adequate data security. Keywords- Embedding; Intermediate Significant Bit; Covertcommunication; Steganography.

communication is taking place. The issue of critical importance, in steganography is that, the very presence of a hidden message in the cover medium must be concealed. Such a requirement is not critical in watermarking problems. Security, capacity, imperceptibility and robustness are essential features of a data hiding system. In this paper we assume that adversary is passive, as such emphasis has been given to high hiding capacity coupled with imperceptibility and security. Rest of the paper is organized as follows. Section II provides information related application areas of data hiding. Section III provides an insight into data hiding system features. Section IV differentiates blind and non blind data hiding systems. In section V literature survey regarding high capacity data hiding techniques has been presented. Section VI provides complete description of proposed work. The results obtained in the proposed technique and the comparison with existing ones is presented in section VII. The paper concludes in section VIII. II. DATA HIDING APPLICATIONS Data hiding encompasses both digital watermarking and steganography has been found useful in following areas: A. Copyright Protection: To protect copyrights and assert ownership of a Multimedia content. Generally a signature called watermark is inserted in the medium that is to be protected against copyright infringments. B. Content Authentication:The use of fragile water marks is made for content authentication. Content authentication is used to verify authenticity of a multimedia content and ensure whether same piece has been received that was transmitted at transmitter or a changed version of original piece.

I.

INTRODUCTION

Stegnography is a type of hidden communication that literally means covered writing [1]. It has been derived from Greek words stegano means covert and graphos means to write The rapid development of technology has increa sed the horizon of communication via various wired and wireless networks specially internet. This has resulted in serious challenges pertaining to integrity and security of data being communicated. Stegnography in such a scenario is being used to avert adversaries. The goal of steganography is to hide information inside a harmless cover medium in such a way that it is not possible even to detect that a secret message exists in cover medium [2]. Although encryption has been one of the solutions for information security, but encrypted messages once intercepted, can easily provide clue to the adversary or enemy that some message of importance is being communicated. Steganography, on the other hand, takes opposite approach and attempts to hide all evidence that

ISBN : 978-1-4673-5144-7/12/$31.00 2012 IEEE

2012 - International Conference on Emerging Trends in Science, Engineering and Technology 193 C. Broadcast monitoring: This is used to monitor if the contracted number of commercials were broadcasted in a given time slot or not. This ensures if the commercials were given due air time or not. D. Fingerprinting: This is used to discourage unauthorized duplication and distribution by embedding a distinct watermark (or a fingerprint) in each copy of the data. E. Metadata binding: Metadata refers to side information embedded in an image that can serve many purposes. For example, a business can embed the Web site address for a specific product in a picture that shows an advertisement for that product. F. Covert communication: To transmit information secretly from transmitter to receiver. III. DATA HIDING SYSTEM: ESSENTIAL REQUIREMENTS Data hiding systems are being used for a varied set of applications. Each data hiding application has its own specific requirements. Therefore, there is no set of requirements to be met by all data hiding techniques. Nevertheless, some general requirements that most of the applications are supposed to have are mentioned below: A. Perceptual Transparency: The data hiding algorithm should be such that the information embedded in the cover medium should not deteriorate the perceptual quality of cover data. An embedding procedure is truly imperceptible if humans cannot distinguish the cover medium from the stego-medium [3]. B. Payload: This refers to amount of information that can be embedded in cover medium. Generally speaking payload for stegnographic applications is much higher compared to watermarking systems C. Security: The security of data hiding techniques is governed by Kerckhoffs assumption states which states that one should assume the method used to encrypt the data is known to an unauthorized party and that the security lies in the choice of a key [4, 5, 6]. Hence a data hiding technique is truly secure if knowing the exact algorithms for embedding and extracting the hidden data does not help an unauthorized party to detect the presence of the secret data or remove it. D. Roubustness: A watermark (secret data) should stay in the cover medium regardless of various signal processing operations carried over host medium, which include all hostile attacks that unauthorized parties may attempt. This requirement is referred to as robustness. It is a key requirement for copyright protection and figerprinting applications but less important for applications like steganography where capacity and security are of prime importance. IV. BLIND AND NON BLIND DATA HIDING The extraction algorithm is an integral and important part of a data hiding system. The extraction process in a data hiding system could be blind or non blind depending upon the strategy used to embed data in the cover media. As such in certain data hiding applications the cover media is essential for the data extraction. For example, in some applications, like fingerprinting, copyright protection and data monitoring, extraction algorithms use the original cover medium to find the watermark. This is called non- blind watermarking. In other applications, e.g., copy protection and indexing, covert communication, extraction algorithms do not have access to the original data. As such the extraction process is rendered more difficult. The algorithms of this kind are referred to blind or oblivious detecting algorithms. Blind detection, although is a tough job, but provides an advantage that receiver does not need original cover medium for secret data extraction. V. REVIEW OF LITERATURE

Information hiding is a new and multidisciplinary field that encompasses cryptography, coding theory, information theory and theory of human perception. In the last decade a lot of research attention has been paid in this direction, [7] gives an idea about acceleration of research in this area. Spatial-domain watermarking techniques for image data include [8, 9]. Some of the earliest reported spatial domain techniques embed m-sequences in to the least significant bit (LSB) of the data to provide an effective transparent embedding technique. M-sequences are chosen due to their good correlation properties so that a correlation operation can be used for secret message detection. Furthermore, these techniques are computationally inexpensive to implement. Several spatial domain data hiding techniques for images are proposed in [10]. An adaptive surrounding pixel technique, which utilizes all eight adjacent neighboring pixels, for embedding secret information, so that imperceptibility grows has been developed and is presented in [11]. Pixel intensity based high capacity data embedding method is presented in [12]. The method improves the modified Kekres algorithm which is based on LSB method. The capacity is improved by

ISBN : 978-1-4673-5144-7/12/$31.00 2012 IEEE

2012 - International Conference on Emerging Trends in Science, Engineering and Technology 194 embedding the payload into the low intensity pixels and hence maximum utilization of cover image. Development of spatial domain stegnographic techniques for gray scale images has been reported in [13]. The schemes are reported to have high hiding capacity and good imperceptibility properties. The authors have discussed the key issues pertaining to data hiding in detail. [14] Reports a high capacity data embedding scheme based on average covariance of the Discrete Cosine Coefficients (DCT). The Most significant Bit (MSB) of the payload is embedded into cover image based on average covariance of cover image. A Peak Signal to Noise Ratio (PSNR) of 46.31% for hiding capacity of 12.50% has been reported for test image size 512x512. An Intermediate Significant Bit (ISB) replacement based watermarking technique has been reported and presented in [15]. The authors have reported that embedding information in the intermediate significant bits improves robustness compared to data hidden in least significant bits. [16] Reports a high capacity data hiding technique based on spatial domain. The host image is partitioned into non-over lapping blocks, with each block containing three 3x3 pixels. In every block these pixels receive special treatment, with an aim to decrease the noise and deviations from the original picture values. The authors have reported PSNR values ranging between 27db to 30db for hiding capacity of 20-26.6%. Most of the reported techniques emphasize upon high capacity and are less descriptive about security of payload. In the proposed technique both the aspects capacity and security have been given due consideration. VI. PROPOSED METHODE The block diagram of proposed high capacity stegnographic system using ISB data embedding is shown in Fig. 1. Firstly, Cover image is prepared for data embedding by breaking it into its constituent bit planes as shown in Fig. 2. The philosophy behind data embedding in the proposed system is, more significant the bit plane, lesser the amount of data embedded in it. This philosophy ensures better perceptual qualities of the stego image. As such data to be embedded in the bit planes has been divided into three variable length data vectors of continuously decreasing lengths. The embedding strategy is depicted in Fig.2. As shown the data vector with total length is divided into three variable length data vectors, and of continuously decreasing magnitude. viz: The data is embedded in the first three ISB planes under the control of a private key. In order to thwart the adversary data is not embedded sequentially. The key, which is generated using Pseudo Random Number Generator (PRNG), ensures a highly randomized data embedding in the three given bit planes. This process is depicted in Fig. 3. The embedding process is carried out in data embedder, that outputs an image containing secret data and is generally termed as stego-image. The security of data embedded is a function of Key Length. The used pseudo random number generator (PRNG) is capable of addressing all the locations in first three Intermediate Significant Bit planes where data is to be embedded. As such the PRNG uses 18 bit seed word to generate the key for embedding data.

A. Embedding Strategy and relationship between length of data vectors : The proposed data hiding system,
breaks the data vector to be embedded, in smaller size vectors equal, in number to the number of ISB planes in which data is to be hidden. The lengths of data vectors can be related in several ways. In the implemened technique the data has been broken into three blocks with lengths L1, L2 and L3. This is because data is to be embedded in three ISBs. The relation between the lengths of data blocks is L1= L/2, L2= 3L/8 and L3=L/8; where L is the total length of data vector to be embedded in the cover medium. The data is embedded in the embedder under the control of a private key. B. Extraction Strategy: The embedding algorithm uses private key to embed the data in the ISBs of cover image. The resultant image yeilded by data embedder is called stego image. At the receiving end the the extraction algorithm uses stego-image along with same key as that used at embedder to extract data from the stego-image. Since cover image is not needed for the retrieval of secret data the proposed system falls in the category of blind detection.

Fig. 1: Proposed high capacity data hiding and corresponding blind extraction system

ISBN : 978-1-4673-5144-7/12/$31.00 2012 IEEE

2012 - International Conference on Emerging Trends in Science, Engineering and Technology 195 A. Hiding Capacity (HC): The amount of data that can be embedded in a cover image without deteriorating integrity of the cover image gives an idea about the hiding capacity. It is also referred to as payload. Hiding capacity is represented by bits per pixel (bpp). It is given by (total number message bits/total number of image bits) multiplied by 100. If n and N respectively denote total message bits and image bits the hiding capacity is given by Hiding Capacity (HC) = (n/N)*100
Fig. 2: Data embedding strategy

B. Peak Signal to Noise Ratio (PSNR): It is an important image, objective, quality index. It is actually a measure of quality of image when external data is embedded in it. It gives an idea about how much deterioration has embedding caused to the image. It is represented as Where mse is mean square error and is given by

Fig.3: Data embedding using Random Key

VII.

RESULTS AND ANALYSIS

The requirements of a data hiding system when used for stegnographic purposes are of twofold viz, high hiding capacity and imperceptibility. The two parameters unfortunately are of opposing nature. This is because embedding more and more data in a cover image results in deterioration of the cover image quality, and as such an adversary could easily guess that some data has been embedded in the cover image. Keeping in view these conflicting features a reasonable amount of data has been taken to be embedded in the cover medium so as to keep degradation in the image quality minimum. For the testing the efficacy of the proposed scheme a set of nine standard grey scale test images (512 x512) were used. Table 1 shows all the used test images with their corresponding stego-image, besides showing hiding capacity (HC) also termed as payload and PSNR in decibels in each case. Table 2 presents details of subset of host test images with amount of percentage data embedded and corresponding peak signal to noise ratio PSNR. In all cases the embedding capacity has been fixed as at 25% that comes out to be (262144x2 bits of payload) except test image lake where the payload of 31.25% has been choosen. Further a comparison of the proposed data hiding scheme with that of Zeki et. al [16] can be seen in table 2. Table 3 shows graphical comparison between the proposed technique and that of Zeki et. al. The hiding Capacity (HC) and PSNR have been calculated as follows.

Where N and M are image dimensions, original and stego images respectively.

and

represent

TABLE I.

VARIOUS TEST IMAGES AND THEIR STEGO VERSIONS

Details of host image, its stego version, PSNR and percentage of data embedded Host Image Stego Image

Boats

Boats, HC=25%, PSNR=37.50db

Peppers

Peppers, HC=25%, PSNR=37.65 db

ISBN : 978-1-4673-5144-7/12/$31.00 2012 IEEE

2012 - International Conference on Emerging Trends in Science, Engineering and Technology 196

Airplane

Airplane, HC=25% PSNR=37.63 db

Baboon

Baboon, HC=25%, PSNR=37.64 db

Lake

Lake, HC=31.25% PSNR=36.08 db Lena Lena, HC=25%, PSNR=37.55 db

TABLE II.

CAPACITY VERSUS PSNR OF VARIOUS TEST IMAGES OF

PROPOSED ALGORITHM

Barbara

Barbara, HC=25% PSNR=37.59 db

TABLE III.

COMPARISON BETWEEN PROPOSED AND ZEKI [16] TECHNIQUE

Milk drop

Milk drop, HC=25% PSNR=37.48 db

Cameraman

Cameraman, HC=25%, PSNR=37.97 db

ISBN : 978-1-4673-5144-7/12/$31.00 2012 IEEE

2012 - International Conference on Emerging Trends in Science, Engineering and Technology 197 VIII. CONCLUSION
[6] Descrambler, Maejo Int. J. Sci. Technolgy. 4(01), 125 -135, 2010. Shabir A. Parah et.al, On the realization of a secure, high capacity data embedding technique using joint top-down and down- top embedding approach Elixir Comp. Sci. & Engg. (49) 10141-10146 , 2012 P. Wah Wong and E. J. Delp, editors. Security and Watermarking of Multimedia Contents II,. Society of Photooptical Instrumentation Engineers, volume 3971, 2000 . R.G. Schyndel, A. Tirkel, and C. Osborne, A digital watermark, in Proc.IEEE Int. Conf. Image Processing (ICIP), pp. 86-90, 1994. R.B. Wolfgang and E.J. Delp, A watermark for digital images, in IEEEProc. Int. Conf. Image Processing, Lausanne, Switzerland, Sept. vol. 3, pp. 219-222. 1996. W. Bender, D. Gruhl, and N. Morimoto, Techniques for data hiding, IBM Syst. J., vol. 35, nos. 3-4, pp. 313-336, 1996. A. Masoud and I. Subariah Adaptive Steganography scheme using More Surrounding pixels, International Conference on Computer Design and Applications, 225-229. 2010. Mehdi Hussain and Mureed Hussain, Pixel Intensity Based High Capacity Data Embedding Method, International Conference on Information and Emerging Technologies,15, 2010. N. I. Wu and M. S. Hwang, Data Hiding: Current Status and Key Issues, International Journal of Network Security, 4(1), pp.1-9. 2010. N. Sathisha et. al, Embedding Information In DCT Coefficients Based On Average Covariance International Journal of Engineering Science and Technology (IJEST), 3 (4), 3184-3194. 2011. M. A Zeki and A. A. Manaf, A Novel Digital Watermarking Technique Based on ISB (Intermediate Significant Bit) , World academy of science Engineering and Technology, vol. 50,989996, 2009 M. A. Zeki et. al, High watermarking capacity based on spatial domain technique Information technology journal:, 10(7),13671373, 2011.

A secure and high capacity data hiding technique with blind detection is presented in this paper. The image in which the data is embedded has been broken into its constituent bit planes. The data to be embedded in the cover medium has been divided into three variable length data vectors. The data vectors are subsequently embedded in first three ISB planes using a private key generated by Pseudo Random Number Generator (PNRG). The PRNG not only embeds data pseudo randomly in various bit planes but it also ensures pseudorandom embedding of data at various pixel locations, thus providing an adequate security to the data carried by the cover image. The technique has been implemented using MATLAB 7. The proposed technique, on an average provides about 8db improvement in PSNR when compared with [16] even when payload is increased on an average by 2.2%. The results clearly show that the proposed technique has a better performance. REFERENCES
[1] C. Cachin, An information-theoretic model for steganography, Information and Computation, Vol. 192, 41-56. 2004 [2] N. Provos and P. Honeyman, Hide and Seek : an introduction to steganography, Computer journal of IEEE security and privacy magazine, Vol. 2(3), pp. 32-40, 2003. [3] M.D. Swanson, M. Kobayashi, and A.H. Tewfik, Mul timedia data-embedding and watermarking technologies, Proc. IEEE, vol. 86, pp. 1064-1087, June 1998. [4] G. M Bhat, Shabir A. Parah et.al, VHDL Modeling and Simulation of Data Scrambler and Descrambler for Secure Data Communication, Indian Journal of Science and Technology, Vol 2, No. 10, pp. 41-43, 2009. [5] G. M Bhat, Shabir A. Parah et.al, FPGA Implementation of Novel Complex PN Code Generator based data Scrambler and

[7] [8] [9] [10] [11] [12]

[13] [14]

[15]

[16]

ISBN : 978-1-4673-5144-7/12/$31.00 2012 IEEE