2010 International Conference on Recent Trends in Information, Telecommunication and Computing

A Novel Approach to Develop a Secure Image based Steganographic Model using Integer Wavelet Transform
Souvik Bhattacharyya
Department of Computer Science and Engineering, University Institute of Technology, The University of Burdwan,Burdwan,India souvik.bha@gmail.com

Avinash Prasad Kshitij

National Institute of Science Technology and Development Studies , New Delhi, India. pdavinash@gmail.com

Gautom Sanyal
Department of Computer Science and Engineering, National Institute of Technology Durgapur, India. gs_cse@nitdgp.ac.in

Abstract--Steganography is a process that involves hiding a message in an appropriate carrier like image or audio. It is of Greek origin and means "Covered or hidden writing". The carrier can be sent to a receiver without any one except the authenticated receiver only knows existence of the information. Steganography is an emerging area which is used for secured data transmission over any public media. Considerable amount of work has been carried out by different researchers on steganography. In this paper, a specific image based steganography technique for communicating information more securely between two locations is proposed. The author incorporated the idea of secret key for authentication at both ends in order to achieve high level of security. Before the embedding operation the cover image is segmented in different objects through normalized cut. As a further improvement of security level, the information has been permuted, encoded through integer wavelet transformation by lifting scheme and segmented in different parts and then finally each part has been embedded through modified lsb embedding method on different cuts of the cover image to form different stego objects. Finally stego image is formed by combining different stego objects and transmit to the receiver side. At the receiver side different reverse operation has been carried out to get back the original information. The proposed embedding approach works by selecting the embedding pixels using some mathematical function and then finds the 8 neighborhood of the selected pixel and map each bit of the secret message in each of the neighbor pixel in a specified manner. Before embedding a checking has been done to find out whether the selected pixel or its neighbor lies at the boundary of the image or not. Keywords: Stego Image, Normalized Cut, Integer Wavelet Transform ,Modified LSB

I. INTRODUCTION The term steganography is not new today. In fact several examples from the times of ancient Greece are available in Kahn [3]. In recent years, everything is trending toward digitalization and with the rapid development of the Internet technologies, digital media can be transmitted conveniently over the network. Therefore, messages need to be transmitted secretly through the digital media by using the steganography
978-0-7695-3975-1/10 $25.00 2010 IEEE DOI 10.1109/ITC.2010.68 173

techniques. Steganography differs from cryptography in the sense that where cryptography focuses on keeping the contents of a message secret, steganography focuses on keeping the existence of a message secret [6, 15]. Another form of information hiding is digital watermarking, which is the process that embeds data called a watermark, tag or label into a multimedia object such that watermark can be detected or extracted later to make an assertion about the object. The object may be an image, audio, video or text only [9]. Although steganography is an ancient subject, the modern formulation of it comes from the prisoners problem proposed by Simmons [1], where two prisoners named Alice and Bob wish to communicate in secret to hatch an escape plan. All of their communication passes through a warden named Eve who will throw them solitary confinement if she suspects any type of secret communication. So they must find out someway of hiding their secret message which gives the birth of steganography. The warden is free to examine all communication exchanged between Alice and Bob can either be active or passive. An active warden will try to alter the communication with the suspected hidden information deliberately in order to remove the information where as a passive warden takes the note of covered communication, informs the others and allows the message to pass through. An assumption can be made based on this model is that if both the sender and receiver share some common secret information then the corresponding steganography protocol is known as then the secret key steganography where as pure steganography means that there is none prior information shared by sender and receiver. If the public key of the receiver is known to the sender, the steganographic protocol is called public key steganography [2, 5]. For a more thorough knowledge of steganography methodology the reader may see [7, 15]. Although all digital file formats can be used for steganography, but the image and audio files are more suitable because of their high degree of redundancy [15]. Fig. 1 shows the different categories of file formats that can be used for

Authorized licensed use limited to: Reva Institute of Tehnology and Management. Downloaded on June 25,2010 at 06:24:58 UTC from IEEE Xplore. Restrictions apply.

steganography techniques. A block diagram of a generic image steganographic system is given in Fig. 2.

FIGURE 1: STEGANOGRAPHY CATEGORY

has been divided into several segments using normalized cut [4, 10] . Each segment of the encoded message has been embedded at each segments of the cover image through modified lsb method to form the stego objects. Stego image will be formed combining the stego objects. This work proposes a novel algorithm with higher security features so that the embedded message can not be hacked by unauthorized user.

A message is embedded in a digital image (cover image) through an embedding algorithm, with the help of a secret key. The resulting stego image is transmitted over a channel to the receiver where it is processed by the extraction algorithm using the same key. During transmission the stego image, it can be monitored by unauthenticated viewers who will only notice the transmission of an image without discovering the existence of the hidden message. Kevin Curran et al [13] propose an image based steganography methods where he describes a set of steganography methods along with their respective merits and demerits. The most common and simplest image embedding method is the least significant bit (LSB) insertion. The LSB insertion embeds the message in the least significant bit of some selected pixels of the Cover image. R.Chadramouli et al. [12] gives an analysis of lsb based steganography techniques. The embedding capacity of LSB method can be increased by using two or more least significant bits. At the same time, not only the risk of making the embedded message statistically detectable increases but also the image fidelity degrades. Hence a variable-sized LSB embedding scheme is presented in [11], in which the number of LSBs used for message embedding /extracting depends on the local characteristics of the pixel. The advantages of LSBbased method are easy to implement. Unfortunately, the hidden message is assailable due to a slight modification from the active warden. Marvel et al. [8] present an image steganographic method, entitled spread spectrum image steganography (SSIS) that hides and recovers the message within digital imagery. The SSIS incorporated the use of errorcontrol codes to correct the large number of bit errors. In recent years many image steganography models have been proposed where the main objective is to protect the transmitted data against any odd. Although increasing the security level of the hidden message of the transmitted data is still an open issue. Silvia Torres Maya et al. [14] presents a steganographical algorithm based on bit plane complexity segmentation, which permits to implement hiding information into images for its sure transmission through a non secure channel. Steganographic algorithm with high security features has been presented in [16-18]. In this paper a specific secret-key image based steganographic model has been proposed which uses an image as the cover data and the secret information is embedded in the cover to form the stego image. Before embedding the secret information has been permuted, encoded through integer wavelet transform by lifting scheme [19-22]. The cover image

FIGURE 2: GENERIC IMAGE STEGANOGRAPHIC SYSTEM

This paper has been organized as following sections: - Section II describes the proposed model and the solution methodology. Mathematical Analysis of some process has been discussed in Section III. Section IV describes different algorithms for different processes used at both at sender side and receiver side. Experimental results are shown in Section V. Section VI contains the analysis of the results and Section VII draws the conclusion. II. DESCRIPTION OF THE PROPOSED MODEL Fig. 3 shows the block diagram of the proposed secret-key image steganographic model. The input messages can be in any digital form, and are often treated as a bit stream. The input message is first permuted and then encrypted using integer wavelet transformation. This encrypted message generates the secret key, which may be called a message enabled key. Before the embedding operation the cover image is segmented in different objects through normalized cut. For the improvement of security level, the encrypted message will be segmented in different parts according to the no of cuts and each part will be embedded through modified lsb method on different cuts of the cover image to form different stego objects. Stego image is formed by combining different stego objects and transmit to the receiver side. At the receiver side, first the stego image will be tested for a specific feature. If that feature matches, the extraction process starts by extracting the cuts of the stego image for formation of the stego objects. Next the embedded message segments are extracted from the different stego objects but in encrypted form. After reassembling the extracted message goes through the decryption via inverse integer wavelet transformation and repermutation and finally the receiver may be able to see the embedded message with the help of same secret key generated at the sender side.

174

Authorized licensed use limited to: Reva Institute of Tehnology and Management. Downloaded on June 25,2010 at 06:24:58 UTC from IEEE Xplore. Restrictions apply.

A. Solution Methodology The proposed system consists of following two windows, one at the SENDER SIDE and the other at the RECEIVER SIDE. The user will be someone who is familiar with the process of information hiding and will have the knowledge of steganography systems. Integer wavelet transformation through Haar will be included as an option prior to steganography for generation of the secret key. The user should be able to select a plaintext message from a file, an image to be used as the carrier (cover image) and then use the embedding method which will hide the encrypted message in the selected carrier image and will form the stego image. Besides this segmented objects extraction of the cover image through normalized cut has been carried out at the sender side before embedding and objects extraction of the stego image through normalized cut at the receiver side before extraction respectively. The user at the receiver side should be able to extract the message from the stego image with the help of different reverse process in sequential manner to un-hide the message from the stego image.

interval or on an irregular grid, or even on a sphere..The wavelet lifting scheme is a method for decomposing wavelet transform into a set of stages. An advantage of lifting scheme is that they do not require temporary storage in the calculation steps and have required less no of computation steps. The lifting procedure consists of three phases, namely, (i) split phase, (ii) predict phase and (iii) update phase.

FIGURE 5: LIFTING SCHEME FORWARD WAVELET TRANSFORMATION

Splitting: Split the signal x into even samples and odd samples:

Prediction Predict the odd samples using linear interpolation: Update: Update the even samples to preserve the mean value of the samples: The output from the s channel provides a low pass filtered version of the input where as the output from the d channel provides the high pass filtered version of the input. The inverse transformed is obtained by reversing the order and the sign of the operations performed in the forward transform.

x even: si x2i. x odd: di x2i+1.

di di {( si + si+1 )/2}. si si + {(di-1+di)/4}.

FIGURE 3: PROPOSED SECRET KEY STEGANOGRAPHY MODEL

FIGURE 6: LIFTING SCHEME INVERSE WAVELET TRANSFORMATION

B. Lifting Scheme Haar Transform In the lifting scheme version of the Haar transform, the prediction step predicts that the odd element will be equal to the even element. The difference between the predicted value (the even element) and the actual value of the odd element replaces the odd element. For the forward transform iteration j and element i, the new odd element, j+1,i would be In the lifting scheme version of the Haar transform the update step replaces an even element with the average of the even/odd pair (e.g., the even element si and its odd successor, si+1):

oddj+1,i = oddj,i - evenj,i

FIGURE 4: GUI BASED STEGANOGRAPHY SYSTEM

III. MATHEMATICAL ANALYSIS OF THE PROCESSES A. Integer Wavelet Transform through lifting The lifting scheme is an algorithm to calculate wavelet transforms in an efficient way. It is also a generic method to create so-called second-generation wavelets. They are much more flexible and can be used to define wavelet basis on an

even j +1,i =

even j ,i + odd j ,i
2

The original value of the oddj,i element has been replaced by the difference between this element and its even predecessor. Simple algebra lets us recover the original value:

175

Authorized licensed use limited to: Reva Institute of Tehnology and Management. Downloaded on June 25,2010 at 06:24:58 UTC from IEEE Xplore. Restrictions apply.

Substituting this into the average, we get even j ,i + even j ,i + odd even j +1,i = 2

oddj,i = evenj,i + oddj+1,i

C. Algorithm for NCUT

j +1,i

even j +1,i = even j ,i

odd j +1,i +
2

C. Extraction of cuts through Normalized Cut Shi and Malik [4,10] proposed the Normalized Cuts algorithm for solving image segmentation problem, which is treated as a graph partitioning problem. The normalized cut criterion measures both the total dissimilarity between the different groups as well as the total similarity within the groups. An efficient computational technique based on a generalized eigen value problem can be used to optimize this criterion. Algorithm of Normalized Cut: A Graph G = (V, E) can be partitioned into two disjoint sets, A and B.

Define the weight function. Prepare the weighted graph G = (V, E) and compute weight matrix using the weight function. Compute (D-W)x=Dx for eigenvectors with smallest eigen values ,where W is the cost matrix and D is the diagonal matrix containing the sum of the cost from node i. Use the eigenvector with second smallest value for bi-partitioning the graph. If the segmented subdivisions are re-partition able then go to step 2. Prepare the NCUT mask, where only the cuts are defined, from the segmented image. Mark each region in the mask with a unique number. Set each pixel in the mask to the region number where it belongs to. For each region in the mask, if any pixel in the mask is within the region then keep the same pixel in the original image and set all other pixels in the image to zero. Repeat step 4 till all the regions are covered. Select a pixel and find its 8 neighbors in counter clockwise direction. Check whether the selected pixel or any of its neighbors is edge pixel or not. If edge pixels then choose a new pixel with neighbors otherwise continue. Form the lsb array starting from the pixel at POS-1 to POS-8 (As shown in figure 8). Add the message bit stream with the lsb array to form a new array. Form the stego image by embedding each bit of the newly formed array in to the next bit of lsb of the pixel from POS1 to POS8. Compare bit wise lsb array with newly generated array after message embedding. If there is match in bit position map it as 0 in message bit stream otherwise map it as 1 in message bit stream.

D. Algorithm for dividing the regions and cut extraction

FIGURE 7: CUTS IN A GRAPH

The degree of dissimilarity between these two pieces can be computed as

E. Algorithm for Message Embedding

The normalized cut (NCUT) is defined as

where

IV. ALGORITHMS In this section, algorithms for different processes used both in the sender side and receiver side are discussed.
A. Extraction of cuts of the Cover Image Segmentation and cut extraction of the cover image is done through combining normalized cut and region growing method. B. Algorithm for Stego Objects Formation

F. Algorithm for Message Extracting

Select the Cover Image. Extract the cuts of the Cover Image through NCUT method. Segment the encoded message based on the no of cuts. Convert the compressed and segmented input text message in binary form and embed into the cuts of the Cover image to form the Stego objects.

G. Algorithm for Pixel Selection One important point needs to be kept in mind that a specific order for selecting the neighbor pixels has to be maintained for embedding / mapping process and also for the process of extraction other wise it would not be possible for retrieve the data in proper sequence. This sequence has been shown in Figure 8.

176

Authorized licensed use limited to: Reva Institute of Tehnology and Management. Downloaded on June 25,2010 at 06:24:58 UTC from IEEE Xplore. Restrictions apply.

3 4 2 1 6 8

After permutation and encryption through integer wavelet transformation: [fAn@CljArqu@i4CeFqrBrACg4ien%AG CA )F B )

B. Original Message Vs Integer Wavelet Transformed Message structure:

FIGURE 8: SEQUENCE OF DATA EMBEDDING

H. Algorithm for the GUI In this section the two algorithimic approach is discussed one for the function of the Sender Side and another for the Receiver Side. o o Sender side
FIGURE 9.ORIGINAL MSG VS ENCRYPTED MSG

Select the Cover Image from the set of images. Check whether the image is in true color (24 bits) or in range of gray, if not error. Select the message in text form. Permute and encode the message through lifting scheme (Haar Transform) and also generate the secret key. Extract the cuts of the Cover Image through NCUT method. Segment the encoded message based on the no of cuts. Embed the encrypted message in the cuts to generate the Stego objects. Assemble the Stego objects to generate the Stego Image.
Receiver side

C.

NCuts of the Cover Image:

Fig. 10(A) and Fig. 10(B) shows the cover image and the stego image. Fig. 10(C) shows basic cut structure of the cover image. Fig 10(A) - (F) shows some cuts of the cover image.

A A

B B

CC

FIGURE 10. A) COVER IMAGE. B) STEGO IMAGE C) CUT STRUCTURE OF THE COVER IMAGE.

A A

B B

Compare the Stego Image with the original for a specific feature and if matches continue. Extract the cuts of the Stego Image through NCUT method. Extract the encrypt form of message segments from the Stego objects. Reassemble the message segments. Decrypt the message with the help of the previous mentioned secret key and reverse transform through lifting scheme. Re-permute the message.

FIGURE 11. ( A) - (F) SOME CUTS OF THE COVER IMAGE

D.

Cover Image vs Stego Image:

V. EXPERIMENTAL RESULTS This section presents the obtained results via different processes mentioned in the proposed model.
A. Results of applying different processes over the message Original Message:- India is a great country and I feel proud to be an Indian.

48 60 60

34
42

43 37 45

52

FIGURE 12: A SEGMENT OF COVER IMAGE WITH SELECTED PIXEL

177

Authorized licensed use limited to: Reva Institute of Tehnology and Management. Downloaded on June 25,2010 at 06:24:58 UTC from IEEE Xplore. Restrictions apply.

REFERENCES 50 60 60 32
42

43 37 45

[1] [2] [3] [4] [5] [6]

54

FIGURE 13: A SEGMENT OF STEGO IMAGE AFTER EMBEDDING 11001001

VI. ANALYSIS OF THE RESULTS In the previous work made by different researchers it has been seen that emphasis is given on message embedding technique to make it robust over any image processing operation. In this work an attempt has been made to increase the level of security of the steganography model by incorporating the idea of secret key, along with the use of encoded and segmented form of the original message. Further the object extraction of the cover image, formation of stego objects, assembly of stego objects to generate stego image and feature matching of stego image has also been used to increase the level of security. The Levels of Security incorporated in the proposed model: Generation of the encrypted form of the secret message through integer wavelet transformation. Generation of secret key. Embedding and extraction through modified lsb. Segmentation of the encrypted message. Extraction of cuts of the cover image and generation of stego objects. Assembly of stego objects to form the stego image. Feature matching of the stego image. All the processes both in sender side and receiver side must be executed in proper sequence. VII. CONCLUDING REMARKS

[7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

[17]

In this paper authors have used the combination of, Integer wavelet transformation through Haar, Embedding through modified LSB and NCUT based Image Segmentation technique on raw images to obtain secure stego-image. The integer wavelet transformation has been used to generate the encrypted form of the message in order to achieve high level of security. The encrypted form of the message is embedded into the cuts of the cover image to obtain the stego-objects. A new and efficient steganographic method for embedding secret messages into images without producing any major changes has been proposed here through modified LSB method. This property enables the method to avoid steganalysis. The integrated approach of combining Integer Wavelet Transformation, Modified LSB and segmentation techniques through NCUT enable secure transfer of the message compared to earlier techniques.

[18] [19] [20]

Gustavus J. Simmons, "The Prisoners' Problem and the Subliminal Channel", in Proceedings of CRYPTO '83, pp 51-67. Plenum Press (1984). Stretching the Limits of Steganography", RJ Anderson, in Information Hiding, Springer Lecture Notes in Computer Science v 1174 (1996) pp 39-48 D. Kahn, The Codebreakers - the comprehensive history of secret communication from ancient times to the Internet, Scribner, New York (1996). J. Shi and J. Malik, "Normalized Cuts and Image Segmentation," Int. Conf. Computer Vision and Pattern Recognition, San Juan, Puerto Rico, June 1997. Scott Craver, "On Public-key Steganography in the Presence of an Active Warden," in Proceedings of 2nd International Workshop on Information Hiding, April 1998, Portland, Oregon, USA. pp. 355 - 368. Ross J. Anderson and Fabien A.P. Petitcolas, "On the limits of steganography," IEEE Journal on Selected Areas in Communications (JSAC), Special Issue on Copyright & Privacy Protection, vol. 16 no. 4, pp 474-481, May 1998. N. F. Johnson and S. Jajodia, "Steganography: seeing the unseen," IEEE Computer.,Feb., 26-34 (1998). L. M. Marvel, C. G. Boncelet, Jr. and C. T. Retter, "Spread spectrum image steganography," IEEE Trans. on Image Processing, 8(8), 10751083 (1999). Digital Watermarking: A Tutorial Review S.P.Mohanty, 1999. J. Shi and J. Malik, "Normalized cuts and image segmentation.,"IEEE Trans. PAMI, vol. 22, no. 8, pp. 888-905, 2000. Y. K. Lee and L. H. Chen, "High capacity image steganographic model," IEE Proc.-Vision, Image and Signal Processing, 147(3), 288-294 (2000). Analysis of LSB Based Image Steganography Techniques ,R. Chandramouli, Nasir Memon, Proc. IEEE ICIP, 2001. An Evaluation of Image Based Steganography Methods,Kevin Curran, Kran Bailey, International Journal of Digital Evidence,Fall 2003. Silvia Torres Maya,Mariko Nakano and Ruben Vazquez Medina Robust Steganography using Bit Plane Complexity Segmentation 1st International Conferenceon Electrical and Electronics Engineering ,2004. T Mrkel,JHP Eloff and MS Olivier .An Overview of Image Steganography,in proceedings of the fifth annual Information Security South Africa Conference ,2005. Study of Secure Steganography model by Souvik Bhattacharyya and Gautam Sanyal at the proceedings of International Conference on Advanced Computing & Communication Technologies (ICACCT2008) An Image based Steganography model for promoting Global Cyber Security by Souvik Bhattacharyya , Gautam Sanyal and Arka Majumdar at the proceedings of International Conference on Systemics,Cybernetics and Informatics (ICSCI-2009) Implementation and Design of an Image based Steganographic model by Souvik Bhattacharyya and Gautam Sanyal at the proceedings of IEEE International Advance Computing Conference (IACC-2009) R. Calderbank, I. Daubechies, W. Sweldens, and B.-L. Yeo. Wavelet transforms that map integers to integers. Appl. Comput. Harmon. Anal., 5(3):332369, 1998. W. Sweldens. The lifting scheme: A construction of second generation wavelets. SIAM J. Math. Anal.,29(2):511-546, 1997.

[21] R. Calderbank, 1. Daubechies, W. Sweldens, and B.-L. Yeo. Wavelet transforms that map integers to integers. Appl. Comput. Harmon. Anal., 5(3):332369,1998. [22] Integer Wavelet Transforms using the Lifting Scheme,GEERT UYTTERHOEVEN DIRK ROOSE ADHEMAR BULTHEEL, CSCC Proceedings, 1999

178

Authorized licensed use limited to: Reva Institute of Tehnology and Management. Downloaded on June 25,2010 at 06:24:58 UTC from IEEE Xplore. Restrictions apply.