An Improved BPCS Image Steganography In Integer Wavelet Transform Domain Using 4×4 Block Size

DOI : 10.17577/IJERTV1IS8044

Download Full-Text PDF Cite this Publication

Text Only Version

An Improved BPCS Image Steganography In Integer Wavelet Transform Domain Using 4×4 Block Size

Sarita1*, Kamlesh Lakhwani2 ,shilpa choudhary3

1Department of Computer Science, Suresh Gyan Vihar, University, Jaipur 302025, India 2 Department of Computer Science, Suresh Gyan Vihar, University, Jaipur 302025, India 2 Department of Computer Science, Suresh Gyan Vihar, University, Jaipur 302025, India

Abstract

Steganography is the process that apply secret information in a multimedia carrier, and carrier may be image, audio, and video files but digital images are the most popular because of their high frequency on the Internet. For hiding secret information in images, there exists a large variety of steganography techniques. Different applications have different requirements of the Steganography technique used. For example, some applications may require absolute invisibility of the secret information, while others require a larger secret message to be hidden. Steganography are used in current security techniques by the remarkable growth in computational power, the increase in security awareness by, e.g., individuals, groups, agencies, government and through intellectual pursuit.

Major issues in Image steganography are to increase the payload capacity, imperceptibility of secret information in stego image & increasing the robustness against steganalysisImage Steganography in Wavelet transform domain have higher robustness against statistical attacks compared to spatial domain & image steganography in Discrete Cosine Transform domain. Integer wavelet transform avoid the losses in fraction which happened in discrete wavelet transform so increase the stego image quality compared to image stenography using discrete wavelet transform. BPCS image steganography is depends on Characteristics of human eyes that our human vision system can not perceive any shape information or secret information in a very complicated binary pattern. So all complex bit plane of cover image can be embedded with secret bit plane without affecting the image quality

so BPCS method has higher invisibility or robustness against visual attacks & high payload capacity and BPCS image steganography in integer wavelet transform have high robustness statistical attacks also while 4 LSB image steganography have only high capacity but low robustness against visual attacks and statistical attacks.

Keywords- Steganography, Hiding, Integer, Wavelet, BPCS

1. Introduction

cryptography is a the art of writing a secret information to make a message un-understandable for a third party such a way that attacker can not decode the unreadable message and cryptography does not hide the existence of the secret information or communication while in steganography, hiding of guess of stego image is first priority[1].

Anderson proposed a the scheme of Least significant based hiding[2] which it is very easy for implementation but have lowest robustness against statistical attacks. Vijay Kumar and Dinesh Kumar analyzed the imperceptibility in different sub band as cH, cV, CD of Discrete Wavelet Transform (DWT) based image steganography. Experiment & Results shows cD sub-band or diagonal detail coefficients band of DWT gives higher imperceptibility or PSNR compared to other band as cH, cV in DWT based image steganography[3]. Gheorghita Ghinea & Adel Almohammad analysed the advantage & disadvantage of hiding data in coloured & grayscale images and payload capacity & effect of hiding in chrominance component of YCRCB images. According to them image steganography in colour

images have higher robustness against visual attacks compared to data hiding in Gray scale images on same payload capacity[4]. R.O. El. Safy, H.H. Zayed & A. El Dessouki proposed an adaptive data hiding technique using the optimum pixel adjustment algorithm to increase the hiding capacity. In this schemes different number of bits in each integer wavelet coefficient is embedded according to a hiding capacity function in order to maximize the hiding capacity without affecting the invisibility in stego image and this method also minimizes the mean square error by optimum pixel adjustment algorithm. In this paper three different cases of hiding scheme is tested according to requirement by the user according to priority in capacity or invisibility. In decoding side there is no error in the recovered message but stego image have low robustness against statistical attacks such as histogram equalization and JPEG compression [5]. Elham Ghasemi, Jamshid Shanbehzadeh and Bahram ZahirAzami proposed a scheme based on Integer Wavelet Transform and Genetic Algorithm which hide information using a mapping function based on Genetic Algorithm in 8×8 block of wavelet coefficient. According to them Optimal Pixel Adjustment Process and Genetic Algorithm are used to an optimal mapping function to increase the payload capacity with low distortions and to decrease the mean square error respectively but The deficiency of this technique is higher of execution time[6]. Michiharu Niimi, Hideki Noda and Bruce Segee proposed secure BPCS which is more robust against the the visual attack compared to conventional BPCS image steganography in spatial domain. According to them, if hiding of both secret data and conjugation flags into a noisy block using image segmentation and complexity threshold is employed then signature of existing of messages or secret patterns do not visible in the stego-images[7]. Tao Zhang, Zhaohui Li and Peipei Shi concluded in their paper that, in the improved BPCS steganography with dynamic different threshold on different planes for a purpose of higher invisibility & higher robustness against analysis of frequency histogram, the existence of secret information & estimate the embedding threshold value can be analyzed using the partial statistical analysis[8]. Peipei Shi, Zhaohui Li and Tao Zhang proposed a BPCS image steganography scheme based on chaos theory that is more robust against the analysis of existence of secret information with desirable visual imperceptibility & payload capacity[9]. Ramani, Prasad and Varadarajan proposed a the BPCS image steganography using integer wavelet transform in which the secret information embedding is employed in bit planes of integer wavelets Coefficients of sub-

band using the Integer Wavelet Transform. For increasing payload capacity without effecting the imperceptibility of the hidden data, the secret data is hidden in IWT coefficient according to the complexity measured in the Bit-Plane. it is a lossless image steganography which use the IWT and BPCS for high data hiding capacity and high imperceptibility. BPCS takes the advantage of human visual system which cannot recognize changes in complex positions of the image[10]. Julio Lopez, Raul Martinez, Mariko Nakhano and Kazuhiko presented an improvement in the Steganalysis method based on statistical moments of wavelet characteristic function. According to this scheme, This Steganalysis method has a low detection of stego image generated by BPCS steganography and this method proposes to use of support vector machine for classification in place of artificial neural network classifier and it is based on the first three moments of characteristic features of the sub bands with the 3-level Haar Wavelet transformation [11]. The Spatial image steganography methods are most applicable in lossless image format, so these method depends on image format[12]. Transform domain image Steganography involves the image transforms & manipulation of algorithms. These techniques embed secret information in more significant areas of the image, that by it is more robust against visual attacks & statistical attacks[14]. In this approaches the embedded message is not lost in conversion between lossy and lossless compression & mostly methods are independent of the image format.

  1. . Related Work

    1. Wavelet Transform

      Wavelet transformation is a powerful image processing transform operation which is used widely used feature extraction, compression and de-noising. Wavelet transform represents the signals with small waves, called wavelet, of limited durations. It provides examination of the signal both in frequency.

      If (t) L2(R), the basic wavelet, (t) is defined as

      C=|(w)|2w dw

      Where (w) is basic wavelets Fourier Transform, w is circular frequency. The wavelet transform decompose the image into four sub band of different frequency groups. Coefficient of low frequency sub band is called approximate components which represents the characteristics of a image while coefficients of high frequency sub-band called

      detailed components which represents noise and redundancy in a image [14].The two-dimensional wavelet transform is achieved by applying the one- dimensional wavelet transform to the rows and columns of the input image consecutively[15].

    2. Discrete Wavelet Transform

      If O is original image then A, H, V & D is calculated as following which represents the approximation, horizontal, Vertical & Diagonal coefficients of Discrete Wavelet Transform respectively.

      Ai,j = (O2i,2j + O2i+1,2j)/2 Hi,j = O2i,2j+1 – O2i,2j Vi,j = O2i+1,2j – O2i,2j Di,j = O2i+1,2j+1 + O2i,2j

      The Inverse Discrete Wavelet Transform is calculated as following.

      O2i;2j = Ai,j – Hi,j/2 O2i,2j+1 = Ai,j – (Hi,j + 1)/2

      O2i+1,2j = O2i,2j+1 + Vi,j – Hi,j O2i+1,2j+1 = O2i+1,2j Di,j Vi,j

    3. Integer Wavelet Transform

      Integer Wavelet Transform is used to avoid problems with floating point precision of the wavelet filters. The LL sub-band of Integer Wavelet Transform appears to be a close copy with smaller scale of the original image while LL sub-band of DWT is distorted as shown in Fig[2.2.9]. Lifting Scheme is one of the method for calculation integer wavelet transform. The decomposing Haar filter for integer wavelet transformation [16] can be applied as following.

      Si,j= S0,2j +S0,2j+1 Di,j=S0,2j+1 -S0,2j

      The Inverse Integer Wavelet Transform is calculated as following:

      S0,2i= Si,i- Di,i2 S0,2i+1= S1,1+Di,i+1

    4. BPCS Image Steganography.

      Basic concept of BPCS is depends on Characteristics of human eyes that our human vision system can not perceive any shape information or secret information in a very complicated binary pattern. So all complex bit plane of cover image can be embedded with secret bit plane without affecting the image quality.

      Embedding algorithm in bpcs image steganography:

      1. 2D Integer wavelet Transform is applied to cover image matrix I to get wavelet coefficient matrix Iiwt. Wavelet Transform decompose a signal into four Sub Band. LL Sub Band or Approximation Band is a Low frequency wavelet coefficient which are consistent with characteristics of a image. LH, HL, HH Sub Band or Detail component are high frequency wavelet coefficient which contain the edge detail in a signal.

      2. Segment integer wavelet transform coefficient matrix Iiwt into 8×8 blocks.

      3. Secret key can be used to determine the order of selection of blocks for embedding.

      4. For Capacity calculation in a block using BPCS Algo, Convert the each channel of each block into Binary as in table 1.1

        o

        10101111

        00100111

        01010000

        11001010

        10100111

        00011110

        01010000

        10111110

        10100100

        00100110

        01011011

        10100101

        10011011

        00100001

        01010101

        10000110

        Plane8 Plane7 Plane6 Plane5

        1001

        0011

        1100

        0010

        1001

        0010

        1001

        0111

        1001

        0000

        1101

        0010

        1001

        0010

        0100

        1010

        1001

        1100

        1101 1100

        0101

        1101

        1101 1000

        Plane4 Plane3 Plane2 Plane1

        0010 1101 0110 1100

        1000 0011 1001 1110

        Compute the border length in each plane from lsb plane to msb plane. In each bit plane border length is defined as changes in consecutive bits row wise & column wise. Calculate the complexity in each bit plane which is defined the ratio of border length and total possible border length in 8×8 block. the maximum possible border length in 8×8 block is 112.

        C=Total Border Length/Maximum Border Length Maximum Border for 4×4 block=112

      5. Determine the capacity of each block finding its number of bit planes other than Most significant plane possessing a complexity higher than a desired threshold.

      6. Determine appropriate complexity threshold for each channel and find the complex planes which have complexity greater than threshold determined in a that channel.

      7. Mapping of complex planes is embedded in a particular pixel. Secret data is converted in binary & binary complex planes of each cover block is embedded with secret binary planes.

      8. When all channel are embedded, Generate a stego image by computing the inverse integer wavelet transform of embedded wavelet matrix.

      9. Generate the stego image by computing the inverse 2D integer wavelet transform.

wavelet transform is high capacity, high robustness against visual attacks & statistical attacks. I have proposed BPCS image steganography in integer wavelet transform domain using 4×4 block size which have higher payload capacity, higher robustness against visual attacks compared to existing BPCS image steganography in integer wavelet transform domain using 8×8 block size.

The embedding and the extracting algorithms are mentioned as following:

Embedding

  1. 2D Integer wavelet Transform is applied to cover image matrix I to get wavelet coefficient matrix Iiwt. Wavelet Transform decompose a signal into four Sub Band. LL Sub Band or Approximation Band is a Low frequency wavelet coefficient which are consistent with characteristics of a image. LH, HL, HH Sub Band or Detail component are high frequency wavelet coefficient which contain the edge detail in a signal.

  2. Segment integer wavelet transform coefficient matrix Iiwt into 8×8 blocks.

  3. Secret key can be used to determine the order of selection of blocks for embedding.

  4. For Capacity calculation in a block using BPCS Algo, Convert the each channel of each block into Binary as in table 2.1

  1. Proposed Work

    Concepts of Proposed work: Basic concept of BPCS depends on Characteristics of human eyes that our human vision system can not perceive any shape information or secret information in a very complicated binary pattern. So all complex bit plane of cover image can be embedded with secret bit plane without affecting the image quality so BPCS method has higher invisibility or robustness against visual attacks. Image steganography using Wavelet transform have higher robustness against statistical attacks. Integer wavelet transform avoid the losses in fraction. So BPCS image steganography in integer

    o

    10101111

    00100111

    01010000

    /td>

    11001010

    10100111

    00011110

    01010000

    10111110

    10100100

    00100110

    01011011

    10100101

    10011011

    00100001

    01010101

    10000110

    o

    Plane8 Plane7 Plane6 Plane5 1001 0011 1100 0010

    1001

    0010

    1001

    0111

    1001

    0000

    1101

    0010

    1001 0010 0100 1010

    Plane4 Plane3 Plane2 Plane1

    1001

    1100

    1101

    1100

    0101

    1101

    1101

    1000

    0010

    1101

    0110

    1100

    1000

    0011

    1001

    1110

    Compute the border length in each plane from lsb plane to msb plane. In each bit plane border length is defined as changes in consecutive bits row wise & column wise. Calculate the complexity in each bit plane which is defined the ratio of border length and total possible border length in 8×8 block. the maximum possible border length in 8×8 block is 112.

    C=Total Border Length/Maximum Border Length Maximum Border for 4×4 block=112

    1. Determine the capacity of each block finding its number of bit planes other than Most significant plane possessing a complexity higher than a desired threshold.

    2. Determine appropriate complexity threshold for each channel and find the complex planes which have complexity greater than threshold determined in a that channel.

    3. Mapping of complex planes is embedded in a particular pixel. Secret data is converted in binary & binary complex planes of each cover block is embedded with secret binary planes.

    4. When all channel are embedded, Generate a stego image by computing the inverse integer wavelet transform of embedded wavelet matrix.

    5. Generate the stego image by computing the inverse 2D integer wavelet transform. Extracting

  1. Compute the 2D integer wavelet transform Iiwt of the stego image Istego as mentioned in above section 2.2.

  2. Segment integer wavelet transform coefficient matrix Iiwt into 8×8 blocks.

  3. Use Secret key to determine the order of selection of blocks for embedding.

  4. Use particular pixel of block to determine the embedded planes.

  5. Extract the embedded plane of the block and Extract the message bits.

  6. Construct the message from extracted bits.

Steganography in wavelet domain should be in those regions where Human Vision System is less sensitive[10]. For this we can adapt the amount of embedded data in each block of wavelet transform domain with a measure of noisiness in that region. We use the bit-plane complexity segmentation (BPCS) as the measure of noisiness as . Each RGB component of a 24-bit bitmap image is an 8-bit value that changes from 0 to 255. In each color plane, the value zero represents the mentioned indarkest shade of that color, where the brightest shading corresponds to the 255 value. Figure 2 shows a 44 test image with the RGB values shown in Table I. Therefore, the R channel is decomposed as indicated in Table II. Now, the bit plane segmentation, visualized in Figure 3, results in eight binary planes for R channel, as shown in Table III. As a benchmark to measure the amount of noisiness of a bit plane, we use the black and white border image complexity defined by Kawaguchi [8]. Based on the definition, the complexity for a black and white border P (equivalent to our segmented plane) is the ratio of the number of total B-W changes in the plane to its maximum possible value, denoted as (P), where 0<(P)<1.

Following measuring the complexity of each plane, we compare the complexity to a threshold to decide if it is a noisy plane. This threshold is to compromise between capacity and imperceptibility. We segment each channel of wavelet transform representation into 88 blocks with pixel values changing from 0 to

255. For each block, we construct the relevant 8-bit planes and compare the bit plane complexity with threshold from the MSB bit plane to the LSB bit plane. Once the first plane with a complexity higher than the threshold is found, we decide on the number of bits that can be embedded in the block pixels. As an example, we can embed five bits of message in the five LSBs of each pixel of the block, if the fourth plane is the first one with a complexity higher than the threshold. For each RGB channel, the threshold is adjusted adaptively according to:

Cth Cin Cmax (1)

where Cin is the parameter to compromise between capacity and imperceptibility ranging from zero to one, Cmax denotes the maximum complexity in the relevant channel, and Cth is the comparative threshold used for making decision on the planes of that channel.

4 .Experiments & Results

In our experiments 4 different images, mostly in research papers of image steganography, is used for test. All four images are 256×256 bitmap RGB colored image of pixel depth 24 which named F15.bmp, Pepper.bmp, Leena.bmp, Baboon.bmp as shown in figure[5.1-5.4]. The algorithm presented in chapter 4 is implemented with slightly difference in step no 2 with 2×2 block size segmentation & 8×8 size segmentation also which is called 2×2 block BPCS image steganography & 4×4 block BPCS image steganography respectively while proposed work is called 4×4 block BPCS image steganography in this dissertation. In 2×2 block BPCS image steganography & 8×8 block BPCS image steganography, maximum possible border length or maximum possible bit pattern is 4 & 112 while in proposed 4×4 block BPCS image steganography it is

24. Experiments are run with different desired threshold from 0.3 to 0.8. The pay load capacity is shown is result is maximum information which can be replaced with secret information which unit is average bits per pixel & PSNR is calculated on same payload capacity in 2×2 BPCS, 8×8 BPCS & proposed 4×4 BPCS technique which unit is dB.

4.1 Measurement Matrices

Measurement units of Quantity of secret information, imperceptibility & robustness against first order statistical attacks are defined as following.

  1. Payload Capacity: Payload Capacity is defined the part of cover image possible to embed with secret information. It is measured either in average bits per pixel or in percentage of cover image.

  2. Imperceptibility: Quality of stego image after the embedding of secret information for robustness against visual Attacks is called imperceptibility of secret information which is proportional to PSNR (Peak Signal to Noise Ratio) as defined in following equation.

    PSNR = 10*logP2MSE

    MSE = x=1My=1N(Sxy- Cxy)

    Where P:-Max. Value in Cover Image

    Sxy: pixel value at xy position in Stego Image

    Cxy : pixel value at xy position in Cover Image.M & N are the pixels in rows & column of Cover image respectively.

  3. Robustness against statistical attacks: First order Steganalysis can be obtained by analysis of frequency histogram. A normal image have a smooth frequency histogram while in frequency histogram of mostly stego images have peaks & not smooth. For higher robustness stego image should have smooth frequency histogram & minimum deviation from original frequency histogram.

sno

Threshold

F15

Pepper

Leena

Baboon

1

0.3

11.4

11.8

12.1

14.6

2/p>

0.4

14.6

10.6

11.2

14.1

3

0.5

9.1

9.4

10.1

12.9

4

0.6

7.9

8.2

8.9

11.5

5

0.7

5.9

6.1

4.6

7.6

6

0.8

2.2

2.3

0.9

1.9

S.

no

Threshold

F15

Pepper

Leena

Baboon

1

0.3

18.5

18.1

19.5

16.1

2

0.4

19.8

20.1

22.4

16.9

3

0.5

22.5

23.2

25.1

19.1

4

0.6

24.9

26.3

27.5

22.1

5

0.7

29.4

30.5

32.9

25.9

6

0.8

35.9

36.8

40.6

33.8

o

sno

Threshold

F15

Pepper

Leena

Baboon

1

0.3

12.5

12.9

13.3

15.6

2

0.4

11.3

11.7

12.3

14.0

3

0.5

10.2

10.5

11.0

12.8

4

0.6

8.8

9.1

9.7

12.3

5

0.7

6.6

6.6

5.2

8.2

6

0.8

2.5

2.7

1.3

2.9

1

0.3

15.9

15.5

16.9

14.1

2

0.4

17.3

17.8

20.1

14.4

3

0.5

20.1

20.9

22.8

16.7

4

0.6

22.1

23.8

24.9

19.7

5

0.7

26.9

28.1

29.3

23.1

6

0.8

32.1

33.4

37.5

30.7

S.

no

Threshold

F15

Pepper

Leena

Baboon

1

0.3

22.1

21.8

23.1

19.9

2

0.4

23.4

24.1

26.2

20.5

3

0.5

26.2

27.1

28.7

22.8

4

0.6

28.7

22.1

30.9

25.6

5

0.7

32.9

33.9

35.9

28.8

6

0.8

38.8

39.2

42.8

37.4

o 5. Conclusion

sno

Threshold

F15

Pepper

Leena

Baboon

1

0.3

9.8

10.1

10.3

12.8

2

0.4

8.9

9.4

9.6

12.2

3

0.5

7.7

8.3

8.5

11.1

4

0.6

6.8

7.4

7.4

9.9

5

0.7

4.9

5.6

3.8

6.7

6

0.8

6.7

2.1

0.8

1.7

sno

Threshold

F15

Pepper

Leena

Baboon

Imperceptibility, Payload capacity & robustness always issues of image steganography. Experiments are done in different images. The Results of experiments shows that proposed BPCS image steganography technique in integer wavelet transform domain using 4×4 in block size have higher payload capacity, higher robustness against visual attacks compared to BPCS image steganography using 8×8 block size in integer wavelet transform domain. Using the mathematical analysis of example shown in favour of proposed work It is concluded that in BPCS image steganography using low block size (ex. 4×4 block size, 2×2 block size), actual bit planes of complex bit pattern is replaced with secret information while in higher block size method(ex. 8×8 block size) some bit planes which have a complex bit pattern is not used for embedding which reduce the pay load capacity & some planes of non complex bit pattern is used for embedding which reduce the imperceptibility. It is also seen from results that proposed technique is more robust against first order statistical attacks because it have a frequency histogram with less peaks & more smoothness compared to 8×8 block size BPCS method. So the proposed BPCS image steganography using 4×4 block size in integer wavelet transform is more efficient than existing 8×8 block size BPCS image steganography in integer wavelet transform. But 2×2 is not efficient than 4×4 & 8×8 block size image steganography. In 2×2 BPCS 25% pixels values is changed for capacity hiding & while in 8×8 BPCS & In 4×4 BPCS image steganography capacity hiding changes 1.56% of pixels & 6.25% of pixels respectively. So in 2×2 BPCS Mean square error is encountered more than 8×8 BPCS & 4×4 BPCS

method on same payload capacity. So 2×2 BPCS technique have low imperceptibility & low payload capacity.

References

[1].S.B. Shadkhan, "Cryptography: currents status & future trends", International Conference on Information & Communication Technology: from Theory to Applications, IEEE, pp:417-418,2004.

[2].R.Anderson and F. Petitcolas, "On the limits of steganography" International Journal of Selected Areas in Communications,IEEE, Vol. 16, No. 4, pp. 474-481,

1998.

[3]. Vijay Kumar, Dinesh Kumar, "Performance Evaluation of DWT Based Image Steganography", International Conference Advance Computing, IEEE, 2010.

[4].Gheorghita Ghinea, Adel Almohammad, "Image Steganography and Chrominance Components", International Conference on Computer and Information Technology, IEEE, pp:996-1001, 2010.

[5].O.El Safy, H.H. Zayed and A.El Dessouki, "A Adoptive Steganographic Technique based on Integer Wavelet Transform", International Conference on Networking and Media Convergence, IEE, pp:111-117, 2009.

[6].Elham Ghasemi, Bahram ZahirAzami, Jamshid Shanbehzadeh, "A Steganographic method based on Integer Wavelet Transform and Genetic Algorithm", International Conference on Communications and Signal Processing,IEEE, pp:42-44, 2010.

[7]. Michiharu Niimi, Hideki Noda, Bruce Segee, " Robust BPCS Steganography against the Visual Attack ", International Conference on Communications and Signal Processing, IEEE, 2007.

[8]. Tao Zhang, Zhaohui Li, Peipei Shi, "Statistical Analysis Against improved BPCS Steganography", International Conference on Advanced Computer Control, IEEE, pp. 237-240, 2010.

[9]. Peipei Shi Zhaohui Li Tao Zhang, "A technique of improved steganography text based on chaos and BPCS", International Conference on Advanced Computer Control, IEEE, pp. 232-236, 2010.

[10]. Ms. K. Ramani, Dr. E. V. Prasad, Dr. S. Varadarajan, Steganography using BPCS to Intrger Wavelet transformed image", International Journal of Computer Science and Network Security, vol.:7, pp:293-302, 2007.

[11].Julio Lopez, Raul Martinez, Mariko Nakhano and Kazuhiko , "Detection of BPCSSteganography Using SMWCF Steganalysis and SVM", International Symposium on Information Theory and its Applications,IEEE, 2008.

[12]. Abbas Cheddad A., Joan Condell, Kevin Curran, Paul Mc Kevitt, "Digital image steganography: Survey and analysis of current methods", International Journal of Signal Processing, Elsevier pp:727-752, 2010.

[13].Venkatraman S., Abraham A. Paprzycki M., " Significance of Steganography on Data Security ", Proceedings of the International Conference on Information Technology: Coding and Computing, 2004.

[14]. H. Wang, S. Wang, "Cyber warfare: Steganography vs. Steganalysis", Communications of the ACM, ACM, 2004.

[15]. C.K Chan and L. M. cheng "Hiding data in images by simple LSB substitution", pattern recognition, pp. 469- 474, 2004.

[16]. A.I. Hashad, A.S. Madani, A.E.M.A. Wahdan, "A robust steganography technique using discrete cosine transform insertion", Third International Conference on Information and Communications Technology, Enabling Technologies for the New Knowledge Society,IEEE, pp. 255-264, 2005.

[17]. A.R. Calderbank,I Daubechies, w. sweldnens,B. Yeo, " Wavelet transforms that map integers to integers" applied and computational harmonic analysis, vol. 5, pp 332- 369,1998.

[18].G. xuan, J. Zhu, Y. Q. Shi, Z.Ni and W.Su., "Distortionless data hiding based on integer wavelet transform", pp.1646-1648,IEEE, 2002.

[19]. P. Chen,and H.Lin, "A DWT approach for image steganography", International journal of applied science and engineering, pp. 275-290, 2006.

Leave a Reply