Robust Digital Image Watermarking using Gradient Vector Quantization and Denoising using Contourlet

Transcription

1 Robust Digital Image Watermarking using Gradient Vector Quantization and Denoising using Contour 1 I.Kullayamma, 2 V.Harika, 3 K.Radhika 1 Assistant Professor, Department of Electronics and Communication Engineering, SV University, Tirupati 2 M.Tech Student, Department of Electronics and Communication Engineering, SV university, Tirupati 3 M.Tech Student, Department of Electronics and Communication Engineering, SV university, Tirupati 1 Ikusuma96@gmail.com, 2 Vemulaharika412@gmail.com, 3 kradhika.0502@gmail.com Abstract we propose a robust quantization-based image ing scheme, called the gradient direction ing (GDWM), and based on the uniform quantization of the direction of gradient vectors. In GDWM, the bits are embedded by quantizing the angles of significant gradient vectors at multiple wave scales. The proposed scheme has the following advantages: 1) Increased invisibility of the embedded because the is embedded in significant gradient vectors, 2) robustness to amplitude scaling attacks because the is embedded in the angles of the gradient vectors, and 3) increased ing capacity as the scheme uses multiple-scale embedding. To quantize the gradient direction, the DWT coefficients are modified based on the derived relationship between the changes in the coefficients and the change in the gradient direction. This ing technique is more Robust to various sizes of Images.In this paper, we pursue a true two dimensional transform that can capture the intrinsic geometrical structure that is key in visual information. The main challenge in exploring geometry in images comes from the discrete nature of the data. Thus, unlike other approaches, such as curves, that first develops a transform in the continuous domain and then discretizes for sampled data, our approach starts with a discrete-domain construction and then studies its convergence to an expansion in the continuous domain. Specifically, we construct a discrete-domain mutiresolution and multidirectional expansion using non-separable filter banks, in much the same way that waves were derived from filter banks. This construction results in a flexible mutiresolution, local, and directional image expansion using contour segments, and thus it is named the contour transform. The discrete contour transform has a fast iterated filter bank algorithm that requires order N operations for N-pixel images. Finally, we show simulation results using contour denoising method over waves. Index Terms digital ing, gradient direction quantization, denoising, waves, contours, filter banks. I. INTRODUCTION Watermarking approaches can generally be classified into two categories: spread spectrum (SS)-based ing and quantization-based ing. The SS type ing, adding a pseudorandom -like into the host signal, has been shown to be robust to many types of attacks. Based on the distribution of the coefficients in the domain, different types of optimum and locally optimum decoders have been proposed. Many SS-based methods have been developed. Wang et al. used a key dependent randomly generated wave filter bank to embed the. Lahouari et al. proposed a robust ing algorithm based on balanced multiwave transform. Bi et al. proposed a ing scheme based on multiband wave transform and empirical mode decomposition (MWT-EMD. In quantization ing, a set of features extracted from the host signal are quantized so that each bit is represented by a quantized feature value. Kundur and Hatzinakos proposed a fragile ing approach for tamper proofing, where the is embedded by quantizing the DWT coefficients. Chen and Wornell introduced quantization index modulation (QIM) as a class of data-hiding codes, which yields larger ing capacity than SS-based methods. Gonzalez and Balado proposed a quantized projection method that combines QIM and SS. Chen and Lin embedded the by modulating the mean of a set of wave coefficients. Wang and Lin embedded the by quantizing the super trees in the wave domain. Bao and Ma proposed a ing method by quantizing the singular values of the wave coefficients. Kalantari and Ahadi proposed a logarithmic quantization index modulation (LQIM) that leads to more robust and less perceptible s than the conventional QIM. Recently, a QIM-based method, that employs quad-tree decomposition to find the visually significant image regions, has been proposed. Quantization-based ing methods are fragile to amplitude scaling attacks. Such attacks do not usually degrade the quality of the attacked media but may severely increase the bit-error rate (BER). During the last few years, many improved techniques have been proposed to deal with this issue, including the use of pilot signals, the design of amplitude-scale invariant codes, such as Trellis codes, orthogonal dirty paper codes and order-preserving lattice codes. Ourique et al. proposed angle QIM (AQIM), where only the angle of a vector of image features is quantized. Embedding the 35

2 in the vector s angle makes the robust to changes in the vector magnitude, such as amplitude scaling attacks. One promising feature for embedding the using AQIM is the angle of the gradient vectors with large magnitudes, referred to as significant gradient vectors. A few ing methods that rely on the significant wave coefficients (which usually correspond to significant edges), have been proposed. Based on the concept of sub image histogram consistency, Kim and Oh presented a ing method for text document images using edge direction histograms. Since this consistency is only valid for text documents, their method can not be directly applied for ing other media types. To our knowledge, no image ing method based on the directions of the significant gradient vectors has been proposed so far. This paper proposes an image embedding scheme that embeds the using uniform quantization of the direction of the significant gradient vectors obtained at multiple wave scales. The proposed method has several advantages: 1) by embedding the in the direction of the gradient vectors (using angle quantization techniques), the is rendered robust to amplitude scaling attacks 2) by embedding the in the significant gradient vectors, the imperceptibility of the embedded is increased, since the HVS is less sensitive to minor changes in edges and textured areas than in smooth region 3) by embedding the at multiple scales, the ing capacity is enhanced. Traditional redundant multistage gradient estimators, such as the multiscale Sobel estimator, have the problem of interscale dependency. To avoid this problem, we employ DWT to estimate the gradient vectors at different scales. To quantize the gradient direction, we propose the absolute angle quantization index modulation (AAQIM). AAQIM solves the problem of angle discontinuity at θ=π by quantizing the absolute angle value. To quantize the gradient angle, we first derive the relationship between the gradient angle and the DWT coefficients. Thus, to embed the bits, the gradient field that corresponds to each wave scale is obtained. This is illustrated in Fig.1 Fig.1. Illustration of five-level gradient field, obtained from five-level wave decomposition. where each gradient vector g j corresponds to the three wave coefficients, and. The straightforward way to embed the bits is to partition the gradient fields into non overlapping blocks. Uniform vector scrambling increases the gradient magnitude entropy, and thus reduces the probability of finding two vectors with similar magnitudes in each block.image Denoising is an important research area serving as the actual foundation for many applications, such as object recognition, digital entertainment, and remote sensing imaging. A wish list for new image representation comes with mutiresolution, localisation, critical sampling, directionality and anisotropy. The first three are successfully provided by separable waves. While, the last two requires contours. Denoising of image is achieved through contours which will be observed in simulation results. II. ANGLE QUANTIZATION INDEX MODULATION (AQIM) AQIM is an extension of the quantization index modulation (QIM) method. The quantization function, denoted by Q (θ), maps a real angle to a binary number as follows: Q (θ) = 0, if θ/ is even 1, if θ/ is odd Where the positive real number represents the angular quantization step size and [.]Denotes the floor function, where the following rules are used to embed a into an angle θ. If Q (θ) = w, then takes the value of the angle at the center of the sector it lies in. If Q (θ) w, then takes the value of the angle at the center of one of the two adjacent sectors, whichever is closer to θ. III. PROPOSED WATERMARK EMBEDDING METHOD Fig. 2 shows the block diagram of the proposed embedding scheme. The is embedded by changing the value of the angle (the direction) of the gradient vectors. First, the 2-D DWT is applied to the image. At each scale, we obtain the gradient vectors in terms of the horizontal, vertical, and diagonal wave coefficients. To embed the, the values of the DWT coefficients that correspond to the angles of the significant gradient vectors are changed. We now discuss the details in the following subsections. The can be embedded in the gradient magnitude and/or in the gradient direction. One disadvantage of the former option is the sensitivity of the to amplitude scaling attacks. However, the second option, embedding the in the gradient direction, is robust to many types of attacks. The angles of the significant gradient vectors, however, remain almost unchanged. The gradient directions form a robust feature 36

3 of an image since their values are not easily changed unless the image quality is severely degraded. The above properties of the gradient direction makes the inserting both robust and imperceptible. We, therefore propose embedding in the directions of significant gradient vectors of the image. Fig.2. Block diagram of the proposed embedding scheme Fig. 3. Block diagram of the proposed decoding method Embedding the Watermark in the Gradient Angle AQIM is one of the best methods for angle quantization. However, it does not account for the angle discontinuity at θ=π. The discontinuity problem arises when the angle is close to. In the proposed ing method, as shown in, the change in each DWT coefficient is computed in terms of dθ. To address this angle discontinuity issue, we propose the absolute angle quantization index modulation (AAQIM). In AAQIM, instead of quantizing the value of the angle, its absolute value is quantized in the interval θ ϵ [0, ᴨ]. The absolute angle quantization function is defined as follows: Q (θ) = 0, if θ/ is even 1, if θ/ is odd IV.PROPOSED WATERMARK DECODING METHOD The bits are decoded following the reverse encoding steps, as shown in Fig. 3. V. WAVELET VS COUNTERLET DENOISING Efficient representation of visual information lies at the heart of many image processing tasks, including compression, denoising, feature extraction, and inverse problems. Efficiency of a representation refers to the ability to capture significant information about an object of interest using a small description. For one-dimensional piecewise smooth signals, like scan lines of an image, waves have been established as the right tool, because they provide an optimal representation for these signals in a certain sense. In addition, the wave representation is enable to efficient algorithms; in particular it leads to fast transforms and convenient tree data structures. These are the key reasons for the success of waves in many signal processing and communication applications. Separable extension from 1-D bases, waves in 2-D are good at isolating the discontinuities at edge points, but will not see the smoothness along the contours. Contour transform is obtained by combining the Laplacian pyramid with a directional filter. Contour transform provides a flexible multi-resolution, local and directional expansion for images. In Contour transform there are two stages 1) a Laplacian pyramid followed by 2) directional filter bank(dfb). 37

4 Laplacian pyramid mids provide a multi-resolution system while directional filter banks give a directional nature to the Contour transform. The Laplacian pyramid decomposition at each level generates a down-sampled low-pass version of the original and the difference between the original and the prediction resulting in a band-pass image. Bandpass images from the Laplacian pyramid are fed into a DFB so that the directional information can be captured. VI. SIMULATION RESULTS Fig.4. The contour filter Denoising: (a) (b) (c) (d) Fig. 5 (a) Original Image, (b) Watermark, (c) Watermarked Image (d) Extracted Watermark The improvement in approximation by contours based on keeping the most significant coefficients will directly lead to improvements in applications, including compression, denoising, and feature extraction. As an example, for image denoising, speckle, Poisson, Gaussian s are added to ed image. A simple thresholding scheme applied on the contour transform is more effective in removing the than it is for the wave transform. Figure displays a comparison of denoising when applying wave and contour hard-thresholding on the ed lena image. The contour transform is shown to be more effective in recovering smooth contours, both visually as well as in PSNR. (e) (f) (g) Fig.6. (e)speckle Ded Images using Wave and Contour, (f) Poisson Ded Images using Wave and Contour, (g) Gaussian Ded Images using Wave and Contour. VII.TABLES MSE, PSNR Calculations of Watermarked Image for various Watermark Sizes Size of Watermarked Speckle Wave Contour Poisson Wave Contour Gaussian Wave Contour MSE 8* PSNR 8*

5 Size of Watermarked Speckle Wave Contour Poisson Wave Contour Gaussian Wave Contour MSE16* PSNR16* Size of Watermarked Speckle Wave Contour Poisson Wave Contour Gaussian Wave Contour MSE32* PSNR32* Size of Watermarked Speckle Wave Contour Poisson Wave Contour Gaussian Wave Contour MSE64* PSNR64* Size of Watermarked Speckle Wave Contour Poisson Wave Contour Gaussian Wave Contour MSE 128* PSNR128* VIII. CONCLUSION We present a gradient direction quantization-based ing scheme. The proposed method embeds the in the direction (angle) of significant gradient vectors, at multiple wave scales. To embed the in the gradient direction, we find the gradient vector in terms of the wave coefficients in sub bands LH, HL, and HH. The gradient angle is then quantized by modifying the DWT coefficients that correspond to the gradient vector. To embed the in each gradient angle, the absolute angle quantization index modulation (AAQIM) is proposed. To extract the correctly, the decoder should be able to identify the gradient vectors that were ed and the embedding order.to solve these problems, we propose scrambling the positions of the gradient vectors uniformly over the wave transform of the image. Increasing the difference in the magnitude of the ed and the uned vectors was also proposed to help identify the ed vectors correctly. From the above experimental results we observe that this ing technique is more Robust to various sizes of Images. The contour filter bank, that can provide a flexible multiscale and directional decomposition for images. The developed discrete filter bank has a precise connection with the associated continuous-domain contour expansion. This connection is defined via a directional mutiresolution analysis that provides successive refinements at both spatial and directional resolution. With parabolic scaling and sufficient directional vanishing moments, the contour expansion is shown to achieve the optimal approximation rate for piecewise C 2 smooth images with C 2 smooth contours. From the simulation results we observed that contour Transform is preferred. REFERENCES [1] Ehsan Nezhadarya, Z Jane Wang, Rabab Kreidieh Ward, Robust Image Watermarking Based on Multiscale Gradient Direction Quantization, IEEE Transactions On Information Forensics and Security, Vol. 6, No. 4, December 2011, Pages: [2] Minh N. Do, Martin Vetterli The Contour Transform: An Efficient Directional Multiresolution Image Representation, IEEE TRANSACTIONS ON IMAGE PROCESSING, pages1-16 [3] J. Zou, X. Tie, R. K. Ward, and D. Qi, Some novel image scrambling methods based on affine modular matrix transformation, J. Inf. Comput. Sci., vol. 2, no. 1, pp , Mar [4] The MathWorks, Inc., Getting Started with MATLAB. help/techdoc/learn_matlab/learn_ matlab.shtml [5] Rafel C Gonzalez, Richard E Woods, Steven L Eddins, Digital Image Processing using Matlab, 2008, Fourth Impression, Pearson Education. [6] Handling Various Attacks in Image Watermarking, Baisa L Gunjal* and Dr. Suresh N Mali** [7] A Matlab contour toolbox is freely available for download from the Matlab Central ( [8] S.Jayaraman, S. Esakkirajan, T. Veerakumar Digital Image Processing,Published In 2009 by McGraw Hill Education (India) Private Limited. 39