1. A Review on: Spread Spectrum Watermarking Techniques
In the name of GOD
A Review on: Spread Spectrum Watermarking Techniques

2. Spread Spectrum communication Cox method Adaptive watermarking
4/26/2009
Table of contents
Introduction
Spread Spectrum communication
Cox method
Adaptive watermarking
Comparison
Conclusion

3. Spread Spectrum Method

4. The first papers on data hiding appeared in the early 1990s.
Introduction
The first papers on data hiding appeared in the early 1990s.
(LSB) embedding techniques
elementary and non-robust against noise.
The period 1996–1998:
The development of Spread Spectrum method codes
image watermarking (Cox et al. 1997)
Video watermarking (Hartung and Girod 1998 )
more robust and have been used in several commercial products.

5. Spread Spectrum Modulation (SSM)
The watermarking problem is analogous to a communication problem with a jammer.
motivated many researchers to apply techniques from SSM (successful against jammers.)
SSM good for military communication systems (secrecy and robustness due to unauthorized person)
The jamming problem:
Standard radio or TV communication system:
TX sends a signal in a relatively narrow frequency band.
Inappropriate in a communication problem with a jammer.
The jammer would allocate all his power to that particular band of frequencies.

6. Spread Spectrum Modulation (Cont’d)
SSM system:
Allocates secret sequences (with a broad frequency spectrum) to the TX, which sends data by modulating these sequences.
RX demodulates the data using a filter matched to the secret sequences. (The codes used for spreading have low cross correlation values and are unique to every user)
Attacker must spread jamming power over all D dimensions.
Owner knows which N dimensions are important.
the message is spread over a wide frequency BW.
The SNR in every frequency band is small (difficult to detect)

7. Advantages of Spread Spectrum Communication
Resist intentional and unintentional interference.
Can share the same frequency band with other users
Protect the privacy, due to the pseudo random code sequence.

8. Spread Spectrum technique for watermarking
Adopt ideas from spread spectrum communication
The WM message is spread over a wide frequency bandwidth (spectrum of the host image)
Hide a D-dim. signal (information to embed) in an N-dim. space (part of original document), N >> D
The SNR in every frequency band is small (difficult to detect)
RX knows the place of WM concentrate weak signals to high SNR output.
Generate noise like carrier or hopping sequence with cryptographically secure methods (Security )
if parts of the message are removed from several bands, enough information is present in other bands to recover the message
it is difficult to remove the message completely without entirely destroying the cover (robustness)

9. Embedding a Direct-Sequence Spread-Spectrum Watermark

10. Recovering a Direct-Sequence Spread Spectrum Watermark

11. Secure Spread Spectrum Watermarks for Multimedia
The first spread spectrum watermarking method based on DCT
Proposed by Cox.
Cox et al. asserted that in order for a watermark to be robust, it need to be placed in the most significant part of the image.
the watermark will be composed of random numbers drawn from a Gaussian N(0,1) distribution

12. Insertion of the watermark
Insert X (WM) into V (DCT coeffs.) results in V’
3 natural formula for computing V’:
The general form:
Do not provide a solution for how to compute  in order to maximize the robustness of the watermark.
(set =0.1)
How choose N?

13. Informed Watermarking Method of Cox et al.

14. Secure Spread Spectrum Watermarks for Multimedia
Evaluating the similarity of the watermark
The extracted watermark might differ from the original watermark.
one should decide on a threshold T, and compare
Set T to minimize the false positives and false negatives.

15. Reducing False Positive (FP) error Selecting the appropriate threshold
4/26/2009
Threshold Comparison
Reducing False Positive (FP) error
Selecting the appropriate threshold 
Select
Threshold =6

16. Experiments
Watermarked image

17. Compression and cropping attacks

18. Uniqueness of the Watermark

19. Disadvantages
The need to have the original image to be able to detect the watermark.
Since the DCT transform is based on the whole image , the transform does not allow for any local spatial control of the watermark.
Does not provide a maximum use of the human visual system

20. Image Adaptive Watermarking
The image adaptive DCT
based on 8x8 DCT framework (Wolfgang et al.)
Wavelet (Podilchuk and Zheng)
Use Just Noticeable Difference (JND) Matrix.
The JND is derived from image independent frequency sensitivity and image dependent luminance sensitivity and contrast masking.
This assists in determining the maximum amount of watermark signal that can be tolerated.
the goal is to place the maximum strength watermark sequence (Why?).

21. Adaptive Image Watermarking
4/26/2009
Adaptive Image Watermarking
HVS models as image Adaptive weights of watermarks:
2) HVS models as filters of Perceptually important image components:
3) A combination of Previous two methods.

22. Podilchuk-Zeng (P&Z) Method
4/26/2009
Podilchuk-Zeng (P&Z) Method
Watermark Embedding: :
threshold calculated for each level & frequency orientation by Watson for 9.7 Daubechies Filters.

23. Watermark Detection
Watermark detection:
Again use similarity and appropriate threshold which is designed to balance false positives and false negatives.

24. Experiment 1 (uniform image)

25. Experiment 2 (non-uniform image)

26. Podilchuk method (Attacks)
Quality
100 80 60 40 20 10 5
Cropping(to 25%)+ Compression 1
0.9671
0.9300
0.9114
0.8843
N/A
cropping(to 6.25%) +compression
0.9588
0.9200
0.8988
0.8600
Compression
0.9680
0.9120
0.8262
0.7050
0.4720
0.3020

27. SSIS Method

29. Results
watermarked
Original

30. JPEG Attack

31. Noise Attack

32. Spread spectrum communication
Conclusion
Spread spectrum communication
watermarking problem is analogous to a communication problem with a jammer.
the WM message is spread over a wide frequency bandwidth (spectrum of the host image)
Cox method.
Adaptive watermarking

33. References
[1] I. J. Cox, J. Kilian, T. Leighton, T. Shamoon, “Secure spread spectrum watermarking”, in IEEE Transaction on Image Processing, vol
[2] C. I. Podilchuk and W. Zeng, “Image adaptive watermarking using visual models”, IEEE journal on selected areas in communication, vol. 16, No. 4, pp , May 1998.
[3] L. M. Marvel, C. G. Boncelet, C. T. Retter, “Spread Spectrum Image Steganography,” IEEE Trans, on Image process., vol 8, no. 8, Aug
[4] W. Lu, H. T. Lu, F. Chang, “Chaos-Based Spread Spectrum Robust Watermarking in DWT Domain,” in Proc. Int. Conf. on Machine Learning and Cybernetics, Guangzhou, Aug, 2005.

34. Any Question?