INTRODUCTION:

Digital Watermarking is a type of marked which is inserted in a noiseless signal such as image, audio or video data. Watermarking is the process of hiding digital content present in the carrier signal. The hidden content may or may not have relation to the carried signal. The fundamental motivation behind Digital Watermarks is to confirm validness or uprightness of the bearer signal or to demonstrate the personality of its proprietors. It is vital in following copyright encroachments and for banknote validation.

Invisible insertion of the watermark is done in the most significant region of the host image so that tampering of that portion with an intention to remove or destroy will degrade the quality and value of the image [1].

Several digital watermarking schemes have been proposed and based on DCT, DFT, and DWT transformations. In this paper, singular value decomposition (SVD)-based watermarking scheme is proposed. SVD transformation preserves both one-way and non-symmetric properties, usually not obtainable in DCT and DFT transformations [2].DWT is very suitable to identify areas in the cover image where a watermark can be imperceptibly embedded [3].DCT has good energy compaction property [4]. However, one of the main problems and the criticism of the DCT is the blocking effect [5].

Ali et al. [6] proposed a watermarking scheme based on Differential Evolution using DWT and SVD. In the embedding process, the singular vector of selected DWT sub-band of the cover is modified with binary watermark image. The proposed method claimed that it offer the solution for false positive problem as suffer by SVD. The false positive problem present in SVD can be removed by using shuffled SVD (SSVD) as presented in [7].Shuffled SVD enhance the reconstructed image quality by breaking an image into set of ensemble images. The shuffled SVD can be used in place of SVD to remove false positive problem in the proposed method. Mehto et al. [8] proposed a medical image watermarking using DWT and DCT. The watermark image contains patient information is embedded in to the medical cover image. The performance of the method is evaluated for different gain without using any attacks. Visual characteristics of low frequency sub-image of DWT and the ability of DCT to remove correlation between DWT coefficients. By doing this, it improves a blind DCT watermarking algorithm to get a new color image digital watermarking schema based on DWT and DCT [9]. Ganic and Eskicioglu [10] present a hybrid method based on DWT and Singular Value Decomposition (SVD). In the embedding process, the singular values of all DWT cover sub band information are modified with singular values of watermark information.

Commonly present disadvantages in traditional watermarking techniques such as inability to withstand attacks which are absent in SVD based algorithms. They offer a robust method of watermarking with minimum or no distortion. DCT based watermarking techniques offer compression while DWT based compression offer scalability. Thus all the three desirable properties can be utilized to create a new robust watermarking technique in [11].Multiple watermarking based on discrete wavelet transforms (DWT), discrete cosine transform (DCT) and singular value decomposition (SVD) has been proposed for healthcare applications. For identity authentication purpose, the proposed algorithm uses three watermarks in the form of medical Lump image watermark, the doctor signature/identification code and diagnostic information of the patient as the text watermarks. The security of the image watermark is enhanced using Arnold transform before inserting into the cover in [12]. Applying the Arnold transform to the watermark improves the robustness greatly. Compared with the SVD and the DCT+SVD algorithms, the proposed algorithm has stronger robustness and faster speed in embedding and extracting [13].

Classification of Digital Watermarking:

Digital watermarking can be categorized as visible watermarking and invisible watermarking. Visible digital watermarking, most well-known is to include private words in the word report; the visible watermarking is equal to a statement. Without obvious computerized watermarking is required in the visual observation can't know about its reality, it is normally hidden inserted into the carrier, when the need to be extracted from the carrier. Unless there is a unique proclamation, the question of investigation of computerized picture watermarking is undetectable watermark. Digital watermarking can be categorized as meaningful watermark and unmeaning watermark. Meaningful watermark refers to the watermark itself which contains specific information effectively namely the two value watermark image, gray image, and the time and place or product serial number. Digital watermarking techniques are categorized into various ways.

Robustness:

Digital watermark is termed as robust when it opposes an assigned class of alterations i.e. the inserted data is identified dependably from the marked signal, even if it is degraded by many changes. Examples of image degradations are JPEG compression, rotation, cropping, additive noise and quantization.

Perceptibility:

Digital watermark is termed as imperceptible if the watermarked content and the original un-watermarked content are perceptually similar i.e. not differentiable. For Example: Digital On-screen Graphics like a Network Logo, Content Bug, Codes, and Opaque images.

Application of DWT, DCT and SVD based watermarking:

Discrete Wavelet Transform:

A discrete wavelet transform (DWT) is any wavelet transform for which the wavelets are discretely sampled. DWT transforms a discrete time signal to a discrete wavelet representation. When compared to other wavelet transforms, the main advantage it has over Fourier transforms is temporal resolution i.e. it captures both frequency and location information.

The DWT is given by

For j≥j₀ and the Inverse DWT (IDWT) is given by:

Discrete Cosine Transform (DCT):

A discrete cosine transform (DCT) is defined as a finite sequence of data points in terms of a sum of cosine functions oscillating at various frequencies. Cosine is used rather than sine functions which are critical for compression; since it describes that fewer cosine functions are needed to approximate a typical signal, whereas for differential equations the cosines express a particular choice of boundary conditions. Particularly, a DCT is a Fourier-related change like the discrete Fourier change (DFT), however uses simply authentic numbers. The DCTs are generally related to Fourier Series coefficients of an at times and symmetrically created progression however DFTs are related to Fourier Series coefficients of a discontinuously expanded course of action in a manner of speaking. DCTs are proportionate to DFTs which is generally twofold the length, working on genuine information with even symmetry in light of the fact that the Fourier change of a genuine and even function is real and even.

DCT Equation:

The DCT equation computes the i, j^th entry of the DCT of a 2-D image as:

Arnold Transform:

Arnold change was proposed by a Russian mathematician V. I. Arnold in the exploration of ergodic hypothesis, which is likewise called as catmapping The change is a procedure of section and joining that realigns the pixel network of advanced picture With the expansion in security issue of picture data as the foundation, some more properties of the time of Arnold change of 2-D were examined by methods for presenting a whole number succession. Arnold change is connected generally in advanced picture scrambling because of its periodicity. The innovation of the picture scrambling is to change a given computerized picture to a scattered one and let its actual articulation of data can't be natural, regardless of the possibility that the estimation of all conceivable mix of conditions will take a high cost. In any case, on the off chance that we become more acquainted with the parameters utilized as a part of the strategy for scrambling, we can simply utilize the opposite scrambling to re establish the first picture. Along these lines, the scrambling cannot exclusively be utilized as a strategy for picture encryption additionally as a method for further pre-handling picture data covering up. A two dimensions Arnold transforms is shown as follows:

where x, y (0.1.2.....N-1) are the coordinates of the pixel. N is the height or width of the square image processed. x^'and y^'are the coordinate of the scrambled image. The transform modifies the position of two pixels, and on the off chance that it is done a few times, a disarranged picture can be created.

Pixels move with periodicity. So, whenever the values change, it creates a totally different Arnold cat map. After multipling for few times, the correlation among the pixels will be totally chaotic. To retrieve the original image, periodicity is required. Assume that the scrambling has performed n emphases, and then one can recover the first picture by performing (T-n) cycles, where T is known as the time of the change. The picture scrambling outline is appeared by performing Arnold Transform with various cycles.

Fig.1: Original image 256*256

Fig.2: Scrambled for 1 time

Fig.3: Scrambled for 8 times

Fig.4: Scrambled for 26 times

Embedding and Extraction Algorithm for image watermark:

The digital image through a 2-D DWT, the sub image can obtain four space of similar size: low frequency approximation LL1, horizontal high-frequency detail HL1, vertical high-frequency detail LH1, and diagonal high-frequency detail HH1 sub images respectively. If the sub image low-frequency approximation LL1 again for 2-D DWT can get two level wavelet transform image of original image and goes on. High-frequency detail sub images of HH obtained by wavelet transform contains the main image texture background and other details of the original image, human visual perception sensitivity to these information is insignificant, in this portion of the information cannot attain better visibility, but this portion of the information is unprotected to interference, thus it cannot promise the robustness of watermarking.

In this approach, watermarking of restorative pictures implants watermarks in the shape picture into the cover picture. Plainly watermarks which contain imperative data and requires more power are implanted into more elevated amount DWT sub-groups. LH1 (Low-high recurrence band) of the primary level DWT is changed by utilizing DCT and after that SVD is connected to the given DCT coefficients. The picture watermark which is additionally changed by utilizing DCT and SVD. The particular estimations of the picture watermark data are inserted into the solitary estimation of the cover restorative picture. Promote, Lump picture watermark is mixed by utilizing Arnold change before installing into the medicinal cover picture which gives the additional level of security. Results are calculated by modifying the gain factor and the various approaches can resist known attacks. This proposed algorithm has two segments, the embedding and extraction process. while the sub image low-frequency approximation LL which has the concentration of most of the energy of the original image will not be modified, so the watermark is inserted into the low-frequency approximation to get better robustness, but the human visual perception is sensitive to this part of the information, thus the watermark information specifically in the embedded portion cannot promise the imperceptibility of the watermark.

Embedding approach of image watermark:

· Third-level DWT is connected to cover picture to disintegrate the picture into comparing sub-groups and LH1 sub-band is chosen.

· Presently DCT is connected to the chosen sub-band and after that SVD is connected to the changed DCT

A_c=U_cS_cV_c^T

Coefficients to get the comparing three matrices U, S and V.

· Encode the Lump watermark image utilizing Arnold Transform

· Apply DCT on encoded Lump watermark picture and then apply SVD to DCT coefficients to get relating matrices like step 2.

A_w=U_wS_wV_w^T

· Modify the cover image singular values of LH1 sub band with the Lump singular values. .

S_wat= S_c + k*S_w

where k is the scaling factor with which watermark pictures are inserted into host image.

· Apply Inverse SVD and obtain modified DCT coefficients by using given equations.

A_wat = U_c*S_wat*V_c^T

· Apply Inverse DCT to modified DCT coefficients and obtain modified LH1 sub band.

· Modify cover image LH1 sub band with the modified LH1 sub band and apply Inverse DWT to obtain the watermarked image.

· To check the robustness of the proposed algorithm, apply attacks and noise to the watermarked image.

Fig.5: Block diagram for embedding process

Extraction approach of image watermark:

· Third-level DWT is connected to cover picture to disintegrate the picture into comparing sub-groups and LH1 sub-band is chosen.

· Presently DCT is connected to the chosen sub-band and after that SVD is connected to the changed DCT coefficients to get the comparing three matrices U, S and V.

A_c=U_cS_cV_c^T

· Apply DCT on encoded Lump watermark picture and then apply SVD to DCT coefficients to get relating matrices like step 2

A_w=U_wS_wV_w^T

· Apply stage 1, stage 2 to watermarked picture to acquire its relating SVD matrices for LH1 sub band.

A_wat=U_watS_watV_wat^T

· Get the Lump singular values from the LH1 sub band singular values of watermarked image and cover image respectively by using following equation:

S_w* = (S_wat-S_c)/k

· Apply inverse Singular Value Decomposition (ISVD) and then inverse Discrete Cosine Transform (IDCT) to obtain extracted watermark.

A_ew= U_w* S_w* V_w^T

· Apply inverse Arnold Transform to decrypt the extracted watermark and obtain final extracted Lump image watermark.

Fig.6: Block diagram for extraction process

EXPERIMENTAL RESULTS:

Fig.7: Cover Image

Fig.8: Watermarked Image

Fig.9: Lump image

Fig.10: Histogram of Lump image

Fig.11: Extracted Lump watermark

Fig.12: Histogram of Extracted Lump watermark

Fig.13: Lump (Cameraman)

Fig.14: Histogram of Lump(Cameraman)

Fig.15: Extracted Lump (cameraman)

Fig.16: Histogram of Extracted Lump (cameraman)

Fig.17: Lump (Lena)

Fig.18: Histogram of Lump (Lena)

Fig.19: Extracted Lump (Lena)

Fig.20: Histogram of Extracted Lump (Lena)

Table 1: Comparison of DICOM cover image with different Lump images.

Cover Image	Lump image	RMSE	PSNR
DICOM	MEDICAL LUMP	17.8313	71.3059
DICOM	CAMERAMAN	43.9817	63.4642
DICOM	LENA	42.1646	63.8306

Table 2: Comparison of different DICOM Cover images with the Lump watermark.

Cover Image	Lump image	RMSE	PSNR
DICOM	MEDICAL LUMP	17.8313	71.3059
CAMERAMAN	MEDICAL LUMP	10.5625	75.8543
LENA	MEDICAL LUMP	9.0634	77.1838

From the figures 9-16 Lump watermark, Extracted watermark and Histogram of those images are shown. The Histogram of Lump watermark and Extracted watermark are same which illustrates that before applying algorithm to the lump watermark and after applying algorithm to the lump watermark, the image has not changed. This means that this algorithm is robust.

From the table 1 it is evident that for DICOM cover image and medical lump watermark has lower RMSE values and higher PSNR values compared to the Lena and Cameraman lump watermark images.

From the table 2 it is clear that Lena cover image has lower RMSE values and higher PSNR values compared to the Cameraman and DICOM cover images. It is because the size of the medical lump used for LENA cover image is 102 x 102 whereas the size of the medical lump used for Cameraman and DICOM cover images is 128 x 128 and 256 x 256 respectively. As the size of the medical lump watermark is less, it resulted in lower RMSE values and higher PSNR values.

Moreover, the recommended technique for wavelet based picture watermarking can be reached out for their application to video watermarking. We might want to additionally enhance the execution which will be accounted for in future correspondence.

ACKNOWLEDGEMENT:

The authors are grateful to the authorities of Sastra University, Thanjavur for the facilities.

CONFLICT OF INTEREST:

The authors declare no conflict of interest.

REFERENCES:

1. Mohanty, Saraju P., and Bharat K. Bhargava. "Invisible watermarking based on creation and robust insertion-extraction of image adaptive watermarks" ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM) 5.2 (2008): 12.

2. Chang, Chin-Chen, Piyu Tsai, and Chia-Chen Lin. "SVD-based digital image watermarking scheme." Pattern Recognition Letters 26.10 (2005): 1577-1586.

3. Barni M, Bartolini F (2001) Improved wavelet-based watermarking through pixel-wise masking. IEEE TransImage Process 10(5):783–791

4. Ahmed KA, Ahmad HA, Gaydecki P (2009) A blind block based DCT watermarking technique for gray level images using one dimensional Walsh coding. International conference on Current Trends in Information Technology, Dubai, pp 1–6

5. Zeng B (1999) Reduction of blocking effect in DCT-coded images using zero-masking techniques. Signal Process 79(2):205–211

6. Ali M, WookAhn C, Siarry P (2014) Differential evolution algorithm for the selection of optimal scaling factors in image watermarking, Special issue on Advances in Evolutionary Optimization Based Image Processing. Eng Appl Artif Intell 31:15–26

7. Guo J-M, Prasetyo H (2014) False-positive-free SVD-based image watermarking. J Vis Commun Image Represent 25(5):1149–1163

8. Mehto A, Mehra N (2016) Adaptive lossless medical image watermarking algorithm based on DCT and DWT. Proced Comput Sci 78:88–94

9. Zhao, Mingwei, and Yanzhong Dang. "Color image copyright protection digital watermarking algorithm based on DWT and DCT." Wireless Communications, Networking and Mobile Computing, 2008. WiCOM'08. 4th International Conference on. IEEE, 2008.

10. Ganic E, Eskicioglu AM (2004) Robust DWT-SVD domain image watermarking: embedding data in all frequencies. Proceedings of the 2004 Workshop on Multimedia and Security, ACM, pp 166–174

11. Navas, K. A., et al. "Dwt-dct-svd based watermarking." Communication Systems Software and Middleware and Workshops, 2008. COMSWARE 2008. 3rd International Conference on. IEEE, 2008.

12. Zear, Aditi, Amit Kumar Singh, and Pardeep Kumar. "A proposed secures multiple watermarking techniques based on DWT, DCT and SVD for application in medicine." Multimedia Tools and Applications (2016): 1-20.

13. Wang, Ben, et al. "An image watermarking algorithm based on DWT DCT and SVD." Network Infrastructure and Digital Content, 2009. IC-NIDC 2009. IEEE International Conference on. IEEE, 2009.

Received on 19.05.2017 Modified on 21.06.2017

Research J. Pharm. and Tech. 2017; 10(9): 2887-2894.

DOI: 10.5958/0974-360X.2017.00510.8