International Journal of Scientific & Engineering Research Volume 2, Issue 10, Oct-2011 1

ISSN 2229-5518

Performance Comparison of Semifragile Watermarking Methods for Image Authentication Archana Tiwari , Dr. Manisha Sharma

Abstract- Data authentication is one of the primary requisites in present day communication systems. In image processing, data authentication is implemented by using watermarking techniques. The specific interest in semifragile watermarking algorithms arises from the multitude of practical and commercial applications, where content needs to be strictly protected, but the exact representa- tion during exchange and storage need not be guaranteed. The alterations on the documents can occur unintentionally or can be implanted intentionally. Semifragile watermarks are more robust and less sensitive to classical user modifications such as JPEG compression, content-preserving operations and content altering manipulations while sensitive to content integrity verification. In present paper the performance of seven semifragile watermarking methods are compared in terms of their PSNR, robustness and temper sensitivity properties.

Key words: semi-fragile watermarking, PSNR, Robustness, Tamper detection.

—————————— ——————————

1 INTRODUCTION

To facilitate the authentication and content- integrity verification for multimedia applications where content-preserving operations are a com- mon practice, semi-fragile watermarking scheme have been proposed in the last few years [1]. This class of watermarks is intended to be fragile only when the manipulations on the watermarked me- dia are deemed malicious by the schemes [2]. Semi- fragile watermarking is a potential solution to the image content authentication problem which seeks to verify that the content of the multimedia has not been modified by any of a predefined set of illegi- timate distortions, while allowing modification by legitimate distortions.
To be an effective image authentication system, it must satisfy the following criteria [8]:
a. Sensitivity: The system must be sensitive to ma- licious manipulations.
b. Tolerance: The system must tolerate some loss of information and more generally non-malicious manipulations.
c. Localization of altered regions: The system should be able to locate precisely any malicious alteration made to the image and verify other areas as authentic.
d. Reconstruction of altered regions: The system may need the ability to restore, even partially, al- tered or destroyed regions in order to allow the user to know what the original content of the ma- nipulated areas was.
In this work focus is on different authentication based semifragile watermarking techniques, thus paper is organized as follows: Section 2 discusses methods of comparison used for comparing differ- ent algorithms, Section 3 gives details on attacks and their countermeasures, Section 4 outlines dif- ferent semifragile watermarking algorithms and Sections 5 & 6 presents analysis and conclusion.

2 Methods for Comparison

The semifragile watermarking method should be moderately robust to discriminate between mali- cious manipulations, such as the addition or re- moval of a significant element of the image, and global operations preserving the semantic content of the image. The semifragile watermarking me- thods should be moderately robust to differentiate between malicious and non malicious attacks. However, the line of demarcation between the be- nign and malicious attacks is application and doc- ument dependent [3]. In this comparison for given algorithms values of PSNR(Peak Signal to Noise Ratio), robustness and tamper sensitivity proper- ties are given, no specific application context is
considered so depending on application particular

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 10, Oct-2011 2

ISSN 2229-5518

methods can be used .Analysis of available water- marking scheme is done on basis of-
I. PSNR: High value of PSNR shows the water- marked image has a better quality, the difference between the original image and the watermarked image is imperceptible.
II. Robustness: Robustness depends on the infor- mation capacity of the watermark, the watermark strength / visibility, and the detection statistics (threshold). Robustness is also influenced by the choice of images (size, content, color depth). The minimal required robustness is highly application dependent. It may not make sense to compare techniques intended for different applications.
III. Tamper detection- Tamper detection aims to
monitor modifications on digital documents where a distinction needs to be made between innocent and malicious alterations. When multimedia con- tent is used for legal purposes, medical applica- tions, news reporting, and commercial transac- tions, it is important to ensure that the content was originated from a specific source and that it had not been changed, manipulated or falsified. This can be achieved by embedding a watermark in the data.
IV. Recovery of content-Lossless recovery of con- tent especially reversible recovery is taken into consideration.
The main content-altering manipulations that must
generate tamper alarm, hence, the non permissible alterations, are the following: Image forgeries in- tended to remove, substitute, or insert objects in the image. Image manipulations that modify the geometry of objects such as their rotation, flipping, translation, and scaling or image manipulations that change the appearance of objects such as color, shade, shadow manipulation, etc.

3 Attacks and Countermeasures

The aim of malicious attacks to authentication sys- tem is not to eliminate the watermarks, but to inva- lidate them-
[I]The attacker may bring a known valid water- mark from a watermarked image as the mark for
another. Then the detector regards it as authentic.
This type of attack can also be performed on the same image: the watermark is first removed, then the image is modified, and finally the watermarkis embedded again. One can use the content-based or reversible watermarking algorithm to resist the attack.
[II] The attacker can modify the marked image without affecting the embedded mark. For exam- ple, if the watermark is embedded in the LSB bit plane of an image, the attacker may attempt to modify the image without disturbing any LSBs. One can choose the transform domain authentica- tion scheme, which has higher security than spatial one. In many quantization-based authentication algorithms, the value of extracted watermark is determined by the quantization interval of marked image's coefficient. If the attacker knows the quan- tization interval, he can modify the image coeffi- cients without changing the value of extracted wa- termark. To solve the problem better, the HVS model can be used to' determine the quantization interval.
[III] The attacker may attempt to completely de- stroy the mark by adding random noise, which maybe the most common type of attack. Improving the robustness of semi-fragile watermark is the only measure so far, but fortunately, if the attacker adds excessive noise, the image's quality will de- crease too much and lost the commercial value consequently [4].
Table 1 discusses different characteristics of semi- fragile watermarking methods and their attack counteracting capacity.

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 10, Oct-2011 3

ISSN 2229-5518

Table 1: Performance analysis of semifragile watermarking methods

Author

Robust- ness

Tamper

Detection

Content

Recovery

Attacks

Guo Rui

Feng

21%

Yes 10×10 block

…….

JPEG compression

Anthony T S

90%

Yes 10×10 block

Possible

Cutting, pasting, noise attacks, JPEG compression

Hyunho keng

60%

Yes 8×8 block

……

JPEG compression, Sharpening, Low pass filtering

Median filtering, Salt Pepper noise, Gaussian noise

Histogram equalization,

Xiaoping

Liang

50%

Yes 3×3 window

Reversible

JPEG compression, Gaussian noise, filtering Content altering , content preserving operations,

Xiaoyun Wu

40%

Yes 4×4 block

Reversible

JPEG compression, Cutting, Displacement

Zhu Xian

40%

Yes

Possible

JPEG compression, Cutting, pepper noise, Gaussian noise, displacement.

Ching Yu

Yang

……..

Yes 4×4 water- mark

Reversible

JPEG 2000, JPEG inverting, brightness

4 Semifragile Watermarking Algorithms

1. Guo rui Feng [2005]:- The keys of the scheme are how to devise the algorithm of permutation and maintenance of the security. In order to achieve these goals, chaotic sequences to permute the host image are suggested [5].

2. Anthony T. S. [2005]:- Watermark is embedded into pinned field which contains texture information of original image. This important properly provides the scheme with special sensitivity to any texture alteration to biomedical image [6].

3. Hyunho King [2006]: - Watermark is inserted in discrete wavelet transform domain and linear correla- tion is used to detect the presence or absence of the watermark on a block by block basis. Each block of

attacked image is divided into low and high frequen- cy coefficient [7].

4. Xiaoping Liang [2007]: - New RSAW (Reversible semi-fragile authentication watermark) scheme is proposed by explaining some characteristics of IWT and multi level embedding. This scheme provides tamper discerning in spatial region from frequency region and tamper localization [8].

5. Xiasyan Wu [2007]: Original image is prepro- cessed by histogram modification; four level IWT is performed on preprocessed image. A binary, watermark is embedded in LLG sub band for tam- per localization and recovery information are em- bedded in high frequency of IWT domain[9].

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 10, Oct-2011 4

ISSN 2229-5518

6. Zhu Xian [2008]:- Copyright information is firstly made chaos by Arnold transform and then self adaptive embedded into host image in differ- ent intensity in wavelet transform domain based on calculating the adjacent wavelet coefficient by exploiting characteristics of Human visual system [10].
7. Ching Yu Yang [2009]: Effective lossless data hiding algorithm based on integer wavelet trans- form, a secret message is embedded in the three high subbands of IWT domain by the proposed
coefficient-bias algorithm [11].
Table 2 shows the values of PSNR of the images recovered after the authenticated image have been JPEG compressed with respect to the image ex- tracted when the authenticated images have un- dergone no compression as given in various algo- rithms.
Table 2 shows the values of PSNR of the images
recovered after the authenticated image have been JPEG compressed with respect to the image ex- tracted when the authenticated images have un- dergone no compression as given in various algo- rithms.
Table 2: PSNR of different images

Author

PSNR OF IMAGES

Author

Baboon

Lena

Boat

Gold hill

Barbara

Pentagon

Pills

Pepper

Camera- man

Pyra- ra- mid

Guo Rui

Feng

37.39 dB

37.40 dB

38.12dB

37.78 dB

36.9dB

38 dB

Anthony T S

38.41 dB

39 dB

37.9 dB

37.8

9 dB

36.9 dB

38 dB

Hyunho keng

41.7 dB

42 dB

40.9 dB

43 dB

41.9 dB

42.2 dB

Xiaoping

Liang

37.38 dB

37.98 dB

38.15 dB

38.35 dB

38.14dB

38.2

1 dB

38.01 dB

33.61 dB

Xiaoyun

Wu

44.48 dB

43.42 dB

43.45 dB

43.46dB

43.45 dB

Zhu Xian

33.83 dB

33.9 dB

34.23 dB

34 dB

34.3 dB

33.61 dB

Ching Yu

Yang

30.64 dB

29.13 dB

29.13 dB

30.12 dB

30.1 dB

5 Analysis And Discussion

This paper presents an investigation of existing semifragile watermarking techniques. Present work is an extension of work done by Ozgur Ekici [3] which covers the research done in the area of semifragile watermarking till year 2004. On com- paring authentication based watermarking me- thods it is observed that, the information carried by the watermark can be accessed using a detec- tion algorithm provided the secret key is known. The various attacks are applied to the images in various methods of semifragile watermarking,
their experimental results show that the algorithm
has robustness resisting conventional compression such as JPEG common image processing opera- tions, such as filtering, lossy compression, noise adding, histogram manipulation, and various geometrical transformations, as a result of achiev- ing a good semi-fragile watermarking algorithm. In some cases it is possible to trade robustness for security; techniques robust to a wider spectrum of image deformations may not have the best robust- ness for specific image deformations. Another im- portant attribute of watermarking is PSNR; high value of it is desirable

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 10, Oct-2011 5

ISSN 2229-5518

for higher quality of image. The tamper detection block size should be flexible, as the attack size can vary from a line consisting of a few tens of pixels long to the entire image plane itself. Tamper proof- ing and hence tampers detection is crucial prob- lem, to know clear demarcation between malicious manipulations and original content of image.

For lossless recovery of images especially reversi-

ble digital watermarking techniques have been individuated so far to be adopted in application scenarios where data authentication and original content recovery were required at the same time[12]. In some applications, it is important that the embedding process be as fast and simple as possible (watermarking images in digital cameras for tamper detection) while the extraction can be more time consuming. In other applications, the speed of extraction is absolutely crucial (e.g., ex- tracting captions from digital video [13].
To develop a fair method for comparing robust-
ness of watermarking techniques, one will have to address the some difficult issues, as listed in table
3.
Table 3 Robustness vs. parameter

Operation Parameter

JPEG compression Quality factor

Blurring Number of opera- tions/kernel

Noise adding Noise amplitude (SNR)

Median filtering Kernel size

Histogram equalization N/A

The following comments can be made on perfor- mance of semifragile method
1. Guo rui Feng- This algorithm preserves effective permuting performance and high security simulta- neously.
2. Anthony T. S. - This algorithm is used for au- thentication of satellite images. Pinned field pro- vide special sensitivity to any texture alteration to the biomedical images.
3. Hyunho King-The objective of this algorithm is
to get a low number of non-detected blocks in non- malicious attacks and a high numbers of non- detected blocks in malicious attacks.
4. Xiaoping Liang- The proposed RSAW scheme is effective in a sense, and is desired in an integrated
and powerful image authentication system, and
can be applied in law, commence, defense and journalism.
5. Xiasyan Wu-This scheme is robust to JPEG lossy compression at a lower quality factor. The embed- ding distortion is small and it guarantees better visual quality of the marked image.
6. Zhu Xian-The algorithm has robustness resisting conventional compression such as JPEG and fragile resisting hostility attacks and falsification, as a re- sult of achieving a good semi-fragile watermarking algorithm.
7. Ching Yu Yang- A semi-fragile reversible data hiding method is, developed which is capable of providing a larger hidden space and a better PSNR while the resulting perceptual quality.

6 Conclusions

In this work seven algorithms are compared on the basis of PSNR, robustness and temper-sensitivity properties. Then, given an application scenario with specific PSNR, robustness and tamper- sensitivity properties, in principle it could be poss- ible to select the adequate algorithm. The compara- tive analysis of semifragile watermarking methods shows that semifragile watermarking is a potential approach for authentication. Semi-fragile water- marking method that can resist content-preserving operations while being sensitive to content-altering manipulations is more practicable for content inte- grity verification.

7 References

[1] R. G. Van Schyndel, A. Z. Tirkel and C. F. Os- borne, “A Digital Watermark”, In Proceedings of IEEE International Conference on Image Processing, Vol. 2, pp 86-90, 1994.
[2] I. J. Cox, Matt Miller, “Electronic Watermarking
The First 50 years”, In Proceeding of IEEE Work- shop on Multimedia Signal Processing, pp. 225 –
230, 2001.
[3] Ozgur Ekici., Bulent Sankur., “Comparative Evaluation of Semi-Fragile Watermarking Algo- rithm”, In Journal of Electronic Imaging, Vol. 13, pp. 209-216, 2004.

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 10, Oct-2011 6

ISSN 2229-5518

[4] Tong , Zheng-ding, “The Survey of Digital Wa- termarking Based Image Authentication Tech- niques”, In Proceedings of IEEE International Con- ference on Signal Processing , pp. 1556-1559, 2002. [5] Guo-rui Feng, Ling-ge Jiang, Chen He, “Permu- tation Based Semi-Fragile Watermark Scheme”, In Journal of IEICE Transaction Fundamentals, Vol. E88–A, pp. 375-378, 2005.
[6] Anthony T. S. Ho, Xunzhan Zhu “A Semi-
Fragile Pinned Sine Transform Watermarking Sys- tem for Content Authentication of Satellite Im- ages”, In Proceeding of IEEE International Confe- rence, pp 737-740, 2005.
[7] Hyunho Kang, Brian Kurkoski, Kazuhiko Ya-
maguchi, Kingo Kobayashi, “Detecting Malicious Attacks Using Semi-Fragile Watermark Based on Visual Model”, In Proceeding of IEEE Internation- al Conference and Computer Society, pp 10-14,
2006.
[8] Xiaoping Liang, Weizhao Liang, Wen Zhang, “.Reversible Semi-Fragile Authentication Water- mark”, In Proceeding of IEEE International Confe- rence on Multimedia and Expo, pp 2122 - 2125,
2007.
[9] Xiaoyun Wu, “Reversible Semifragile Water- marking Based on Histogram Shifting of Integer Wavelet Coefficients”, In Proceedings of IEEE In- ternational Conference on Signal Processing, pp.
501 – 505, 2007.
[10] Zhu Xi'an, “ A Semi-Fragile Digital Water- marking Algorithm in Wavelet Transform Domain Based on Arnold Transform”, In Proceedings of IEEE International Conference on Signal Processing,
pp.2217, 2008.
[11] Chang Min Hwang, Ching Yu Yang, P Yen Chang, Wu-Chih Hu, “A Semifragile Reversible Data Hiding by Coefficient Bias Algorithm”, In Proceedingsf IEEE International Conference on Intelligent Information Hiding and Multimedia Signal Processing, Vol.1, pp. 132 –139, 2009.
[12] D. Zou, Y. Q. Shi, Z. Ni, and W. Su, “A Semi- Fragile Lossless Digital Watermarking Scheme Based on Integer Wavelet Transform,”, In Journal of IEEE on Circuits and Systems for Video Tech- nology, Vol. 16, pp. 1294-1300, 2006.
[13] Z. Ni, Y.-Q. Shi, N. Ansari, and W. Su,
Reversible Data Hiding” In Journal of IEEE on
Circuits and Systems for Video Technology, Vol.
16, pp. 354-361, 2006.

Authors

1.Mrs Archana Tiwari-Associate Professor and HOD Electronics and Instrumentation,Chhatrapati Shivaji Institute of Technology, Durg

Email:archanatiwari@csitdurg.in

2.Dr Manisha Sharma- Professor and HOD Elec- tronics and telecommunication, Bhilai Institute of Technology, Durg

Email: manishasharma1@rediffmail.com

IJSER © 2011 http://www.ijser.org