Conceptual design of edge adaptive steganography scheme based on advanced lsb algorithm

International Journal of Electronics and Communication Engineering & Technology (IJECET), ISSN 0976 –

6464(Print), ISSN 0976 – 6472(Online), Volume 6, Issue 1, January (2015), pp. 39-48 © IAEME

39

CONCEPTUAL DESIGN OF EDGE ADAPTIVE

STEGANOGRAPHY SCHEME BASED ON ADVANCED

LSB ALGORITHM

RAMAKRISHNA HEGDE1, Dr.JAGADEESHA S

2

1Department of Computer Science and Engineering,

SDM Institute of Technology, Ujire, India,

2Department of Electronics and Communications Engineering,

SDM Institute of Technology, Ujire, India,

ABSTRACT

This research article outlines the efficient algorithm in which secrete data are embedded in

edge regions of the image. Here region selection totally depending on size of the secrete message. If

less information to be embedded, then sharp regions in the image are selected for embedding bits. If

the size of secrete information is large, then the more edge regions are released to accommodate all

the secrete bits. This means that edge regions are selected adaptively with the size of the secrete

message. Our work focuses on inserting the bits we used Least Significant Bit Matching Revisited

(LSBMR) scheme and consider the relationship between characteristics of the region and size of the

secrete message. And also this paper proposes to preserve higher visual quality of the stego image.

Keywords: Steganography, Least Significant Bits (LSB), Edge Embedding, Unobtrusiveness,

Security.

I. INTRODUCTION

Steganography is an efficient and well known technique for secure data transmission over the

public networks. The secrete information can be communicated to the other authorized user, by

embedding secrete messages in cover image and it generates stego image. This stego image is

transmitted over the public network and authorized user can extract the secrete data from the stego

image by efficient data extraction algorithm. There are many good algorithms are available for

steganography.

INTERNATIONAL JOURNAL OF ELECTRONICS AND

COMMUNICATION ENGINEERING & TECHNOLOGY (IJECET)

ISSN 0976 – 6464(Print)

ISSN 0976 – 6472(Online)

Volume 6, Issue 1, January (2015), pp. 39-48

© IAEME: http://www.iaeme.com/IJECET.asp

Journal Impact Factor (2015): 7.9817 (Calculated by GISI)

www.jifactor.com

IJECET

© I A E M E



40

Steganography is the art and science of writing hidden messages in such a way that no one,

apart from the sender and intended recipient, suspects the existence of the message. In many new

applications for military and civilian purpose, the contributions of steganography are immense. The

word steganography is of Greek origin and means "concealed writing" from the Greek words

steganos meaning "covered or protected", and graphei meaning writing". The first recorded use of

the term was in 1499 by Johannes rithemius in his Steganographia, a treatise on cryptography and

steganography disguised as a book on magic. As people become aware of the internet day-by-day,

the number of users in the network increases considerably thereby, facing more challenges in terms

of data storage and transmission over the internet, for example information like account number,

password etc. Hence, in order to provide a better security mechanism, we propose a data hiding

technique called steganography along with the technique of encryption-decryption.

Steganography is the art and science of hiding data into different carrier files such as text, audio,

images, video, etc. In cryptography, the secret message that we send may be easily detectable by the

attacker. But in steganography, the secret message is not easily detectable. The persons other than

the sender and receiver are not able to view the secret message.

1.1 Characteristics of Steganography

The steganography should have the following characteristics to function efficiently.

1.1.1Unobtrusiveness - The steganography should be properly invisible or inaudible, or its presence

should not interfere with the media content being protected.

1.1.2 Robustness – The covered data should not be accessed or removed by any other user except

the authorized user. If only a partial knowledge is available, then attempts to remove or destroy a

hidden data should result in severe degradation of the host data before the hidden data is lost. In

particular, the steganography should be robust in the following areas:

1.1.3 Unambiguity - Retrieval of the hidden data should unambiguously identify the owner.

Furthermore, the accuracy of owner identification should degrade gracefully in the face of attack.

1.1.4 Security- The security of steganography techniques can be interpreted in the same way as the

security of encryption techniques Kerchoff’s assumption state that one should assume that the

method used to encrypt the data is known to an unauthorized party and that security must lie in the

choice of a key.

The applications of steganography spread over many areas. Some of the applications are

point to point convert communication to convey secrete information between trusted parties with

host media as camouflage, modern printers, military applications, education, posting secrete

communications on the web to avoid transmission, embedding corrective audio or image data in case

corrosion occurs from a poor connection or transmission. Basic idea of steganography is to hide the

secrete data in master file (carrier file) and communicated to the other user over the networks. Master

files can be of images, audios, text, video file. Most cases images are used as covered media. In

digital steganography, electronic communications may include steganographic coding inside of a

transport layer, such as a document file, image file, program or protocol. Media files are ideal for

steganographic transmission because of their large size. As a simple example, a sender might start

with an innocuous image file and adjust the color of every 100th pixel to correspond to a letter in the

alphabet, a change so subtle that someone not specifically looking for it is unlikely to notice

it.During transmitting the hidden message in covered media over the public network can be detected

by using the method called steganalysis.

II. PREVIOUS WORKS

Lots of research work is done on steganography and some of the work mentioned here. Many

of the applications like copy right protection, content annotation, access control and transaction



41

tracking, the secrete information can be hidden into compressed video bit stream. For instance, [1]

used data hiding techniques to assess the quality of compressed video in the absence of the original

reference. The authors of [2] used data hiding to enable real time scene change detection in

compressed video.

Least Significant Bit (LSB) replacement algorithm is one of the popular and age old

algorithm. In this data hiding scheme, the LSB of master image or covered image is overwritten by

the bit stream of the secrete message according to the pseudorandom number generator (PRNG) . As

a result there is a possibility of introducing some structural changes and for unauthorized users over

public networks become easy to detect the hidden secrete information even though the low

embedding rate by using some of data hiding algorithms such as Chi-squared attack [3],

regular/singular groups (RS) analysis [4], sample pair analysis [5], and the general framework for

structural steganalysis [6], [7]. Most of the LSB replacement algorithms introduces structural

changes. To avoid this up to maximum extent LSB matching (LSBM) technique can be employed. In

LSBM if the secrete bit does not match the LSB of the master image, then +1 or -1 is added

randomly to the corresponding pixel value. Till now several data hiding algorithm [8]-[11] are

proposed to analyze LSBM scheme.

In [8], the author introduced detector using centre of mass (COM) of the histogram

characteristic function (HCF). In [9], the author proved that on a gray scale image, HCF COM does

not work well, so introduced two ways of applying HCF COM method. Namely utilizing the down

sample image and adjacent histogram instead of traditional histogram. In a recent work [11], Li et al.

proposed to calculate calibration-based detectors, such as Calibrated HCF COM, on the difference

image. It proves that Ker’s approach [9] outperformed by new detector and acceptable accuracy at an

embedding rate 50% is achieved. The paper [12] presents an improved data hiding technique based

on BCH (n,k,t) coding. Wang and Moulin [13] have shown that perfect steganography is possible

with zero risk of detection as long as the embedder has perfect knowledge of the cover distribution.

Research papers , Hamming code [14], [15], simplex code [16], binary BCH code [17], Reed–

Solomon code [18], and syndrome-trellis code [19] have been used for secure steganography. Böhme

[20], [21] explain about the security of the steganography from the point of the adversary

assumptions and cover assumptions. The first information-theoretic approach was proposed by

Cachin [22], [23]. He carried out the research on security of a steganographic system using the

Kullback–Leibler (KL) divergence between the distributions of cover and stego media. Binary BCH

codes [24] have been investigated and seem to be good candidates of the first type and the second

type as well. Picture quality and statistical undetectability are two key issues related to

steganography techniques. The research paper [25] proposed closed-loop computing framework that

iteratively searches proper modifications of pixels/coefficients to enhance a base steganographic

scheme with optimized picture quality and higher anti-steganalysis capability.

III METHODOLOGY

First and foremost, during data embedding stage, some of the parameters are initialized. For

region selection in the master image and subsequent processing of this image, those initialized

parameters are used. Abstract data embedding steps involved in this scheme are mentioned in

TABLE 1. Master images which contain more edges are better than the master image having fewer

edges to apply this scheme efficiently. In this scheme, if the selected region is sufficiently enough to

hide all the secrete data, the no need to modify the parameters. Otherwise parameters values should

be modified and reselect the region in the master image. Finally some post processing on embedded

image to obtain the stego image. Otherwise parameters should be revised to reselect the region to

insert the secrete bits until all the secrete bits are inserted.



42

For a different image content and secrete message M, the parameters are also different. These

parameters can be inserted in pre determined region in the cover image. Because these parameters

required during data extraction process at the other end of the network.

Table 1. Various steps involved in data embedding.

Step 1. Input master image

Step 2. Initialize the some parameters.

Step 3. Preprocess the master image by the help of shared keys.

Step 4. Select the region in the master image.

Step 5. If the selected region is sufficient for secrete data hiding, then hide the secrete

data and post process the embedded image to generate stego file(master image +

secrete data).

Step 6. If the selected region is insufficient to hide the secrete message, then modify the

parameters and reselect the region in the master file.

Step 7. Repeat step 5.

The abstract steps involved in data extraction is shown in TABLE 2. Data extraction from the

stego image is more simpler than the data embedding into master image to generate stego image.

During data extraction initially parameter informations are extracted from the stego image and then

does some preprocess on the stego image. Data inserted regions in stego image are identified and by

applying extraction algorithms, secrete data are obtained.

Table 2: Abstract Steps Involved In Data Extraction.

Step 1. Input is Stego image (master image + secrete data).

Step 2. Extract the parameters.

Step 3. Preprocess stego image with the help ofshared keys.

Step 4. Identify the region where data is inserted.

Step 5. Extract the secrete data from the Stego image.

Here we use clear difference between two adjacent pixels as criteria for selecting the regions

in a cover image. LSBMR is used as data hiding technique.

3.1. Data Embedding

• Step 1: The master image of m x n is initially divided in to smaller nonoverlapping blocks Bz

x Bz.pixels. Every small block is rotated with a random degrees (Range 0, 45,

90,135,180,225,270), as determined by the secrete key key1 to improve the security. The rotated



43

small blocks are rearranged as a row vector V by raster scanning. And then vector is divided into

nonoverlaping embedding units with every two consecutive pixels (Xi, Xi+1 ), where i = 1,3 …….,

mn-1.

• Step 2: As per the LSBMR scheme, for each embedding unit, two secrete bits can be

embedded. For a given secrete message M, threshold value T is calculated as follows. Let EU(t) be

the set of pixel pairs whose absolute differences are greater than or equal to parameter t.

EU(t) = {(Xi, Xi+1 )||Xi –Xi+1 | ≥ t, V (Xi, Xi+1 )€ V

Threshold T = argmaxt{2x|EU(t)|≥|M|}

Where t ϵ {0,1……31}, |M| size of the secrete message M, |EU(t)| denotes total

number of elements in the set EU(t).

• Step 3: Performing data hiding on the set of

EU(T) = {(Xi, Xi+1 )||Xi –Xi+1 | ≥ T , V (Xi, Xi+1 )€ V

We deal with the above embedding units in a pseudorandom order determined by a

secret key key2. Here we perform data hiding with following four case.

Case 1. LSB(Xi) = mi & f(Xi, Xi+1) = mi+1

(X’I, X’i+1) = (Xi, Xi+1)

Case 2. LSB(Xi) = mi & f(Xi, Xi+1) ≠ mi+1

(X’I, X’i+1) = (Xi, Xi+1 + r)

Case 3. LSB(Xi) ≠ mi & f(Xi -1, Xi+1) = mi+1

(X’I, X’i+1) = (Xi-1, Xi+1)

Case 4. LSB(Xi) ≠ mi & f(Xi -1, Xi+1) ≠ mi+1

(X’I, X’i+1) = (Xi+1, Xi+1)

Where mi and mi+1 are two secrete bits to be embedded.

• Step 4: After data hiding, resulting region is divided into nonoverlapping Bz x Bz blocks.

These blocks are rotated by the random number of degrees based on key1. Here in this case random

degrees are opposite. The parameters (T,Bz) embed into a predetermined regions in the cover

image.

3.2 Data Extraction

To extract the secrete data from the stego image, we first extract the side information i.e T

and Bz. The stego image is divided into nonoverlapping Bzx Bz blocks. Rotate these blocks into

random degrees with the help of secrete key key1 and rearrange them into the row vector V’. Now we

get the embedding units by dividing the V’into nonoverlapping blocks with two consecutive pixels.

For each qualified embedding unit say (X’i, X’i+1), where | X’i+1 –X’i| ≥ T, we extract two secrete bits

mi and mi+1 as follows.

mi = LSB(X”i), mi+1 = LSB ( [X’i / 2] + X’i+1

IV EXPERIMENTAL RESULTS

4.1 Hiding Capacity and Image quality

In this section, we are presenting the proposed techniques experimental results and compared

with some existing techniques also. Size of the color image we took here is 1024 x 768 or 768 x

1024. These selected color images are converted into gray images and applied our proposed scheme.



44

(a) Cover or master image (b) Embedding rate < 40%

(c) Embedding rate >40%

Fig 1.Images with modified pixels. (a) Cover or Master image

(b) Image with embedding rate < 40% (c) Image with embedding rate >40%

The Fig.1(a) is the cover image or master image. The Fig.1 (b) and (c) shows that positions of

those modified pixels after data hiding by using our proposed technique with embedding rates less

than 40% and greater than 40% respectively in the cover image . The threshold value T is in the

range of 20 to 5 for embedding rates of less than 40% and 3 to 0 for embedding rates of greater than

40%. When the threshold value is T equal to 0, then complete location of the covered image is

available for embedding the secrete bits. Here we observed that, for lower embedding rates, only

sharper edges are used for hiding the secrete data and remain untouched for other clear areas. For

the higher embedding rates, more locations can be released adaptively by decreasing the threshold

value T. For example, when embedding rates are greater than 50% , then the some of clear location

( like clear sky here) in the covered image are released for hiding the secrete bits i.e it all depends on

size of the secrete message.

The Table.3 shows the average PSNR as per the per the paper [26] and our proposed

technique. It observed that quality of LSBMR technique is better in terms average PSNR, since it

uses +_1 embedding scheme and modification rate is also lower. In our proposed scheme average

PSNR is little lower than the LSBMR method because of readjusting the embedding unite to assure

the better data extraction. Overall our proposed method is nearly the best in terms of average PSNR

and average modification rate among other popular techniques



45

Table 3: Average Psnr and the Modification Rate of Stego Images with Different Stegonagraphy

Algorithms and Different Embedding Rates.

4.2 Stego Image Visibility

One of the main characteristics of the better quality steganographic algorithm is, for human

eyes, content of the secrete data into the cover image should not be visible or guessed by the

(a) (b) (c)

Fig 2 . (a) Cover Image (b) Stego Image with 30% embedding rate (c) Stego image with 50%

embedding rate.

Embedding

Rates

Steganography

Alorithm Average PSNR

Avg. Rate of

Modification.

30 %

LSB Based LSBM 56.4 0.1500

LSBMR 57.4 0.1125

Edge Based

PVD 46.8 0.1465

IPVD 49.0 0.1471

HBC 56.4 0.1500

Proposed 56.5 0.1197

50 %

LSB Based LSBM 54.2 0.2500

LSBMR 55.2 0.1875

Edge Based

PVD 44.5 0.2441

IPVD 46.8 0.2452

HBC 54.2 0.2500

Proposed 53.9 0.2033



46

(a) (b)

(c) (d)

(e)

Fig. 3 (a) LSB of cover image and stego images for four different embedding techniques with

embedding rate of 50%. (a) LSB of LSBM stego (b) LSB of LSBMR stego (c) LSB of PVD stego

(d) LSB of IPVD stego (e) LSB of Proposed techniques.

Unauthorized user in the public network even after modifying the pixel value. Visibility of

the original cover image and stego image should remain same and for human eye slight modification

in the pixel value will not be visible. In our proposed technique, although we insert the secrete data

in to the edge region of the cover image, it would not leave any visual artifacts in LSB image.

Fig.2 illustrates the cover image and stego images after applying our proposed technique.

Fig.2 (a) is the cover image or master image, Fig.2 (b) is the stego image with 30% embedding rate,

Fig.2 (c) is the stego image with 50% embedding rate. Here we observed that original cover image

and stego images are all looks similar to the human eye. The stego images with 30% embedding rate

and with 50% embedding rates are looks similar.. Human eyes cannot differentiate both these stego

images.



47

Fig.3 shows the LSB of stego images of different embedding techniques with 50% embedding

rates. It is observed that LSB images for LSBM, LSBMR , PVD and IPVD are almost similar in

nature and when it zoomed, some of artifacts can be visible. Our proposed technique matches with

other embedding techniques and quality of steganography is improved in terms of unobtrusiveness.

V CONCLUSION

In our research work, edge adaptive steganographic technique with advanced LSB algorithm

is used.. As we pointed out in the work that, most of natural images contains the smooth region and

some textual information also. While embedding the secrete bits in those smooth areas and textual

information regions, the structure and visibility of the stego images are bound to change. When stego

image looks differently than the original master or cover image, then unauthorized users can easily

extract the secrete information. So to keep the one of the important characteristic of the

steganography, unobtrusiveness intact and to avoid recognizing the stego image, then cover image

and stego image should not be differentiated by human eye. Our proposed steganographic technique,

edge adaptive with LSBMR achieves this goal. In most of the previous embedding techniques,

pixel/pixel pair selection mainly depending on PSNR without considering the size of the message

and content region characteristics. By implementing above proposed approach, the following results

obtained depending on the size of message. The edge regions are released for insertion of bits based

on size of the secrete message. Our proposed approach achieved 56.5 as average PSNR and .01197

as average rate of modification for 30% embedding rate. For 50% embedding rate 53.9 as average

PSNR and 0.2033 as average rate of modification is achieved.

REFERENCE

1. M. Carli, M. Farais, E. D. Gelasca, R. Tedesco, and A. Neri, “Quality assessment using data

hiding on perceptually important areas,” in Proc.IEEE Int. Conf. Image Processing, ICIP,

Sep. 2005, pp. III-1200-3–III- 1200-3.

2. S. Kapotas and A. Skodras, “A new data hiding scheme for scene change detection in H.264

encoded video sequences,” in Proc. IEEE Int. Conf. Multimedia Expo ICME, Jun. 2008, pp.

277–280.

3. A. Westfeld and A. Pfitzmann, “Attacks on steganographic systems,”in Proc. 3rd Int.

workshop on Information Hiding, 1999, vol. 1768, pp. 61–76.

4. [4] J. Fridrich, M. Goljan, and R. Du, “Detecting LSB teganography in color, and gray-scale

images,” IEEE Multimedia, vol. 8, no. 4, pp. 22–28, Oct. 2001.

5. S. Dumitrescu, X.Wu, and Z.Wang, “Detection of LSB steganography via sample pair

analysis,” IEEE Trans. Signal Process., vol. 51, no. 7, pp. 1995–2007, Jul. 2003.

6. A. D. Ker, “A general framework for structural steganalysis of LSB replacement,” in Proc.

7th Int. Workshop on Information Hiding, 2005, vol. 3427, pp. 296–311.

7. A. D. Ker, “A funsion of maximum likelihood and structural steganalysis,”in Proc. 9th Int.

Workshop on Information Hiding, 2007, vol. 4567, pp. 204–219.

8. J. Harmsen and W. Pearlman, “Steganalysis of additive-noise modelable information hiding,”

Proc. SPIE Electronic Imaging, vol. 5020, pp. 131–142, 2003.

9. A. D. Ker, “Steganalysis of LSB matching in grayscale images,” IEEE Signal Process. Lett.,

vol. 12, no. 6, pp. 441–444, Jun. 2005.

10. F. Huang, B. Li, and J. Huang, “Attack LSB matching steganography by counting alteration

rate of the number of neighbourhood gray levels,” in Proc. IEEE Int. Conf. Image

Processing, Oct. 16–19, 2007, vol. 1, pp. 401–404.



48

11. X. Li, T. Zeng, and B. Yang, “Detecting LSB matching by applying calibration technique for

difference image,” in Proc. 10th ACM Workshop on Multimedia and Security, Oxford, U.K.,

2008, pp. 133–138.

12. Rongyue Zhang, Vasiliy Sachnev, Magnus Bakke Botnan, Hyoung Joong Kim, Member,

IEEE, and Jun Heo, Member, IEEE, “An Efficient Embedder for BCH Coding for

Steganography”, IEEE Trans, Information Theory, Vol 58, No 12, December 2012.

13. Y. Wang and P. Moulin, “Perfectly secure steganography: Capacity, error exponents, and

code constructions,” IEEE Trans. Inf. Theory, vol. 54, no. 6, pp. 2706–2722, Jun. 2008.

14. R. Crandall, Some notes on steganography 1998 [Online]. Available: http://os.inf.tu-

dresden.de/westfeld/crandall.pdf

15. A. Westfeld, “F5: A steganographic algorithm, “High capacity despite better steganalysis”,”

in Information Hiding, ser. Lecture Notes Comput. Sci.. Berlin, Germany: Springer, 2001,

vol. 2137, pp. 289–302.

16. J. Fridrich and D. Soukal, “Matrix embedding for large payloads,” IEEE Trans. Inf.

Forensics Security, vol. 1, no. 3, pp. 390–395, Sep. 2006.

17. D. Schönfeld and A. Winkler, “Embedding with syndrome coding based on BCH codes,” in

Proc. 8th ACM Workshop Multimedia Security, 2006, pp. 214–223.

18. C. Fontaine and F. Galand, “How Reed–Solomon codes can improve steganographic

schemes,” EURASIP J. Inf. Security, vol. 2009, pp. 274845-1–274845-10, 2009.

19. T. Filler, J. Judas, and J. Fridrich, “Minimizing additive distortion in steganography using

syndrome-trellis codes,” IEEE Trans. Inf. Forensics Security, vol. 6, no. 3, pp. 920–935, Sep.

2011.

20. R. Böhme, “An epistemological approach to steganography,” in Information Hiding, ser.

Lecture Notes Comput. Sci.. Berlin, Germany: Springer, 2009, vol. 5806, pp. 15–30.

21. R. Böhme, Advanced Statistical Steganalysis. New York: Springer- Verlag, 2010.

22. C. Cachin, “An information-theoretic model for steganography,” in Information Hiding, ser.

Lecture Notes Comput. Sci.. Berlin, Germany: Springer, 1998, vol. 1525, pp. 306–318.

23. C. Cachin, “An information-theoretic model for steganography,” Inf.Comput., vol. 192, no. 1,

pp. 41–56, 2004.

24. D. Schönfeld and A. Winkler, “Embedding with syndrome coding based on BCH codes,” in

Proc. 8th ACM Workshop Multimedia Security, 2006, pp. 214–223.

25. Guo-Shiang Lin, Yi-Ting Chang, and Wen-Nung Lie, ” A Framework of Enhancing Image

Steganography With Picture Quality Optimization and Anti-Steganalysis Based on Simulated

Annealing Algorithm”, IEEE Trans on Multimedia, Vol 12, No 5, August 2010.

26. S. Pereira, S. Voloshynovskiy, M. Madueno, S. Marchand-Maillet, and T. Pun, “Second

generation benchmarking and application oriented evaluation,” in Proc. 4th Int. Workshop on

Information Hiding, 2001, vol. 2137, pp. 340–353.

27. Geetha C.R. and Dr.Puttamadappa C., “Modified Weighted Embedding Method For Image

Steganography” International journal of Electronics and Communication Engineering

&Technology (IJECET), Volume 4, Issue 3, 2013, pp. 154 - 161, ISSN Print: 0976- 6464,

ISSN Online: 0976 –6472

28. Sonali Patil, Kapil Tajane and Janhavi Sirdeshpande, “Analysing Secure Image Secret

Sharing Schemes Based on Steganography” International journal of Computer Engineering &

Technology (IJCET), Volume 4, Issue 2, 2013, pp. 172 - 178, ISSN Print: 0976 – 6367, ISSN

Online: 0976 – 6375

29. Saurabh Singh and Dr. Ashutosh Datar, “An Efficient Hash Based Steganography Technique

For Text Message Using Color Images” International journal of Computer Engineering &

Technology (IJCET), Volume 4, Issue 4, 2013, pp. 150 - 162, ISSN Print: 0976 – 6367, ISSN

Online: 0976 – 6375