DCT and Simulink Based Realtime Robust Image Watermarking

Durgansh Sharma, Manish Prateek & Tanushyam Chattopadhyay
International Journal of Image Processing (IJIP), Volume (8) : Issue (4) : 2014 214
DCT and Simulink Based Realtime Robust Image Watermarking
Durgansh Sharma durgansh@gmail.com
Ph.D. Student, UPES, Dehradun India
Asst. Professor, Jaipuria Institute of Management, Noida, India
Manish Prateek mprateek@ddn.upes.ac.in
Professor, Centre For Information Technology,
College of Engineering Studies, UPES, Dehradun, India
Tanushyam Chattopadhyay t.chattopadhyay@tcs.com
Senior Scientist, R&D, Innovation Lab,
Tata Consultancy Services, Kolkata, India
Abstract
Ownership of digital content has become a serious matter, due to the exponential raise in the
global repository of digital multimedia content, like images are to be considered in this paper. The
validated proof as an imperceptible and robust watermark is needed to be embedded in the digital
images. This paper proposes a simulation of DCT with Fuzzy Logic based HVS model for
Realtime Robust Image Watermarking technique using Simulink.
Keywords: Realtime, Digital Image Watermarking, Fuzzy logic, HVS, Simulink.
1. INTRODUCTION
Digital image watermarking is a process used for embedding a set of information in an image for
enhancing its authenticity. It has been used through various optimized techniques but mainly on
gray scale images already available in the repository. In this paper, we propose a simulink based
model for realtime robust image watermarking. The robustness of the process of image
watermarking includes DCT and Fuzzy Inference System, which implements HVS (Human Visual
System) for embedding the watermark in an image, captured using webcam in realtime. It
transforms the captured image from spatial domain to frequency domain using DCT and applies
fuzzy logic to implement the HVS logic for embedding the watermark in the host image.
Within a span of few years, we have observed the exponential raise in the usage of digital media
for distribution of images. This model could be extended for the realtime image watermarking in
camera enabled mobile devices to improve the authorization of the image distributed using digital
media which seems to have no boundaries with the open ended environment called internet.
2. REVIEW
The model in this paper proposes the technique to incorporate the authorization watermark in the
host image captured in real time. Motwani et al. [1] used MAMDANI type Fuzzy Inference
System (FIS), its input parameters are derived from Human Visual System (HVS) using the
sensitivity towards brightness, edge and contrast of the gray scale image which has been further
improved to use blue frame of a colour image captured using webcam. Charu et al. [2] further
used and extended the research work using the three layered Fuzzy-BPN having a layer
configuration of (3-3-1) for learning mechanism through 50 iterations. Charu et al. [2] divided an
image of size 256x256 into 1024 blocks of size 8x8 and compute its sensitivity, on the basis of
the variance computed using Fuzzy-BP, the blocks were filtered and the random sequence of
numbers are embedded as watermark. This procedure generated a good quality imperceptible

watermarked image, but, in this paper we try to insert the system identification numbers using the
suggested techniques of Zhao, Jian, and Eckhard Koch [5][6]. Saraju et al.[3] has suggested a
simulink based realtime perceptual watermarking architecture for video broadcasting through
FPGA. Saraju et al. [3] has suggested very useful methods of combining cryptography and
watermarking for better results in hiding useful data in a video frame as a visible transparent
watermark, and Diffie et al. [7] has suggested various new ways of cryptography which can give
better results in data hiding. We try to use his suggestive method in an image captured in a single
shot through webcam using simulink.
3. MODELLING AND ANALYSIS
We consider a 640x480 pixel image captured using webcam for this presented work. The
characteristics are modeled using Eckhard et. al. [5, 6] presented that the multimedia data must
contain a label or code, which could identify it uniquely as property of the copyright holder. The
embedded watermark extracted from the signed image using proposed algorithm was compared
for the similarity correlation using SIM(X, X
*
) proposed by Cox et al. [4]. this parameter is
determined for recovered watermark. Computed values show a good significance level of
optimization in the process of embedding and extraction of watermark.
Sharma et al. [9] proposed Fuzzy inference system (FIS) is used to embed the watermark in the
host image in the DCT domain. The FIS is based on a set of 27 inference rules using SIGMOID
way of interpreting the logical inputs, based on the facts of HVS based sensitivity towards noise in
the image with respect to brightness, texture or contrast, edges.
Rule
No.
Luminance
Sensitivity
Contrast
Sensitivity
Edge
Sensitivity
Weighting
Factor
1 DARK LOW SMALL LEAST
2 DARK MEDIUM SMALL LEAST
3 DARK HIGH SMALL LEAST
4 MEDIUM LOW SMALL LEAST
5 MEDIUM MEDIUM SMALL LEAST
6 MEDIUM HIGH SMALL LEAST
7 BRIGHT LOW SMALL LEAST
8 BRIGHT MEDIUM SMALL LEAST
9 BRIGHT HIGH SMALL LEAST
10 DARK LOW MEDIUM LESS
11 DARK MEDIUM MEDIUM HIGH
12 DARK HIGH MEDIUM HIGHER
13 MEDIUM LOW MEDIUM LESS
14 MEDIUM MEDIUM MEDIUM AVERAGE
15 MEDIUM HIGH MEDIUM AVERAGE
16 BRIGHT LOW MEDIUM LESS
17 BRIGHT MEDIUM MEDIUM AVERAGE
18 BRIGHT HIGH MEDIUM HIGHER
19 DARK LOW LARGE LESS
20 DARK MEDIUM LARGE HIGHER
21 DARK HIGH LARGE HIGHEST

22 MEDIUM LOW LARGE LESS
23 MEDIUM MEDIUM LARGE AVERAGE
24 MEDIUM HIGH LARGE HIGHER
25 BRIGHT LOW LARGE LESS
26 BRIGHT MEDIUM LARGE HIGHER
27 BRIGHT HIGH LARGE HIGHEST
TABLE 1: HVS based 27 Rules for Fuzzy Inference System.
In this paper, we propose a technique to embed imperceptible watermark in an image in realtime.
The model constitutes of webcam needed to acquire an image in realtime, Matlab version 8.0
with Simulink running on a computer. The acquired image constitutes of RGB colour frames; we
propose to extract the Blue frame for embedding the watermark in it and merge it with other two
Red and Green colour frames to reconstitute the image. This process makes the image
watermarking robust and optimized.
The host image captured in realtime in spatial domain having the size of 640x480 pixels is divided
into the blocks of 8x8 pixels each. Discrete Cosine Transformation (DCT) is used for the
transformation of these blocks in the frequency domain. All the three HVS characteristics
mentioned formerly are computed over these blocks as follows:
The Luminance Sensitivity: It is derived from the DC coefficients from the DCT blocks of the
host image according to following formula:
‫ܮ‬௜ =
௑ವ಴,೔
௑ವ಴ಾ
(1)
Where, XDC,i denotes the DC coefficient of the i
th
block and XDCM is the mean value of the DC
coefficients of all the blocks put together.
The Contrast Sensitivity: The contrast sensitivity is derived from the texture content of a region
of 8x8 blocks in an image. The value of variance computed of an image block is provided to the
direct metric for the quantification of the texture as a parameter. A routine proposed by Gonzalez
et. al. [9] is used through MATLAB. The execution of this routine is given by (2).
(2)
where, f is the input image or the sub-image (block) and t is the 7 – element row vector, one of
which is the variance of the block in question.
The Edge Sensitivity: The edge could be detected in an image using the threshold operation;
edge sensitivity can be quantified as a natural effect to the calculation of the block threshold T.
The Matlab image processing toolbox implements bluethresh(f) routine which computes the block
threshold using histogram – based on Otsu’s method of computing graythresh(f) [8]. The
implementation of this routine is given by (3)
(3)
Where, f is the host sub-image (block) in question and T is the computed threshold value. These
three parameters are fed into the FIS shown in Figure 3.
t = statxture(f)
T = bluethresh(f)

FIGURE 1: Simulink Model for DWM.
Once the watermark is embedded then Quality assessment of the signed image is done by
computing Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR).
Extracting Watermark from Signed Image and Computing SIM(X, X
*
) Parameter: Firstly, the DCT
of both host and signed images are computed block wise. Thereafter, the computed coefficients
are subtracted from each other and the watermark is recovered. Let the original and recovered
watermarks be denoted as X and X
*
respectively. A comparison check is performed between X
and X
*
using the similarity correlation parameter given by eq. (1).
ܵ‫ܯܫ‬ሺܺ, ܺ∗ሻ =
∑ ሺ௑,௑∗ሻ೙
೔సభ
∑ ඥሺ௑,௑∗ሻ೙
೔సభ
(1)
4. RESULTS
The profile summary shows the time consumed in the entire process of realtime image
watermarking using the proposed method.

International Journal of Image Processing (IJIP)
FIGURE
Following are the result of watermark embedding process adopted in this paper using
FIGURE 3: Original Image Captured in Realtime
, Manish Prateek & Tanushyam Chattopadhyay
of Image Processing (IJIP), Volume (8) : Issue (4) : 2014
FIGURE 2: Profile Summary of Time Consumed.
Following are the result of watermark embedding process adopted in this paper using
Original Image Captured in Realtime. FIGURE 4: Watermarked Image in Blue
Channel.
MSE: 4.2053dB; PSNR: 41.957dB
218
Following are the result of watermark embedding process adopted in this paper using SIMULINK
Watermarked Image in Blue
dB

5. CONCLUSION
Computed value of SIM(X, X*) parameter for the image depicted in Figure 4 (Singed Image) is
18.5987 which indicates a good watermark recovery process. The time consumed in image
watermarking is computed as approx. 12 seconds, this model could be extended for the realtime
digital image watermarking in camera enabled mobile devices for improving the authenticity of
images captured and shared using Smartphone.
6. REFERENCES
[1] M.C. Motwani, and C.H. Frederick Jr. "Fuzzy Perceptual Watermarking For Ownership
Verification." In IPCV, pp. 321-325. 2009.
[2] C. Agarwal, and A. Mishra. "A Novel Image Watermarking Technique using Fuzzy-BP
Network." In Intelligent Information Hiding and Multimedia Signal Processing (IIH-MSP), 2010
Sixth International Conference on, pp. 102-105. IEEE, 2010.
[3] S.P. Mohanty, and K. Elias "Real-time perceptual watermarking architectures for video
broadcasting." Journal of Systems and Software 84.5 (2011): 724-738.
[4] I.J. Cox, K. Joe, F. T. Leighton, and S. Talal "Secure spread spectrum watermarking for
multimedia." Image Processing, IEEE Transactions on 6, no. 12 (1997): 1673-1687.
[5] J. Zhao, and K. Eckhard "Embedding Robust Labels into Images for Copyright Protection." In
KnowRight, pp. 242-251. 1995.
[6] J. Zhao, and K. Eckhard. "Towards robust and hidden image copyright labeling." In IEEE
Workshop on Nonlinear Signal and Image Processing, pp. 452-455. Neos Marmaras,
Greece, 1995.
[7] W. Diffie and M. Hellman, “New directions in cryptography”, IEEE Transactions on
Information Theory, vol. IT-22, pp. 644-654, 1976.
[8] R.C. Gonzalez, R.E. Woods and S.L. Eddins, Digital Image Processing Using MATLAB,
Pearson Education (2005), pp 406 and 467.
[9] D. Sharma, M. Prateek, T. Chattopadhyay, “Optimized Robust Image Watermarking”,
Proceedings of 4th International Conference on Emerging Trends in Engineering &
Technology, October 25th -27th, 2013, IETET (2013), pp 99-106.

DCT and Simulink Based Realtime Robust Image Watermarking

More Related Content

What's hot (19)

Viewers also liked (18)

Similar to DCT and Simulink Based Realtime Robust Image Watermarking (20)

Recently uploaded (20)

DCT and Simulink Based Realtime Robust Image Watermarking