ISSN: 78 33 Volume, Issue, December 03 A Watermaring Algorithm Based on HT- DWT-TGF for Security Applications C. SUBBA RAO M.E., (PhD) B.V.LAKSHMI (Mtech),ECE Associate Professor, ECE PVP Siddhartha Engineering College. PVP Siddhartha Engineering College Phone No: 990876076, Vijayawada. Kanuru, Vijayawada Phone No: 9704335849, Abstract: - This research paper, scopes a Reversible watermaring scheme for digital images based on DWT & Triangular Generator functions (TGF) and Hardmard Transform. The two ey aspects of Reversible watermaring schemes are copyright protection and robustness. The proposed watermaring algorithm divides the image into sub bands. There are different types of level decompositions such as level,level,level 3.In this design of the project we are using level 3 decomposition. To every pixel of the image three basic techniques (DWT, TGF, and HT) is applied and then XOR operation for the three basic technique is performed. The performance of watermaring algorithm based on pea to signal noise ratio (PSNR), Normalized coefficient (NC), Image Enhancement Factor (IEF) is obtained. The system applies a triangular number generating function to strengthen the binary watermar and employs for watermar synchronization. Experimental results show that the proposed system provides good fidelity of watermared and recovered images and robustness to certain geometrical and non geometrical attacs. Index Terms Discrete Wavelet Transform (DWT), Frequency domain, Hardmard Transform, Reversible Watermaring, Triangular Generating Function. I. INTRODUCTION Today digitization develops day by day, the protection of digital information is important. In order to resist different inds of infringement, a new technology that called watermaring had been put forward to in the international scope. Watermar is sequence carrying information about the copyright owner To embed into the digital image [], audios and videos in order that owners can read it out while unauthorized users cannot easily read it. There are many methods to embed the watermar. It can be divided into two classes: Spatial-domain watermars and transform-domain watermars. The spatial domain is so simple that the watermar can be damaged easily, but the transform- Domain algorithm can be resist intensity attac, watermar information can t be damaged easily. The Transform algorithm includes chiefly DWT, DFT and DCT [, 3, and 4]. Wavelet transform is superior to timefrequency transform for its inner predominance. For example, wavelet has the character of multi-resolution, which can avoid the rectangle brought by DCT. In fact, it has more application fields in engineering and computer science. In this paper, a new reversible watermaring algorithm that embeds a meaningful binary image into the color images is proposed based on HT-DWT- TGF.Some important types of watermaring based on difference watermars [3] are given below: Visible watermars: It is a simple, analogous to stamping a mar on paper. The data is digitally stamped. This is applicable only to images. Example: On television channels visible watermaring is seen when their logo is visibly superimposed in the corner of the screen. All Rights Reserved 03 IJARCET 35
ISSN: 78 33 Volume, Issue, December 03 Invisible watermar: It is a complex concept. It is most often used to copyright data such as author, distributor etc. cover imag e watermar noise attacs Watermar Communicati on Fig. () Digital Watermaring system II. PROBLEM DESCRIPTION Watermar decoder Dected waterm The main purpose of this paper is to contribute in study of comparison of frequency domain techniques are used for image processing as for reservable blind watermaring. Section III describes Basic principle and theoretical part of watermaring. Section IV Describes DWT method for proposed purpose. Section V shows experimental output and comparison result. Section VI concludes the paper in which DCT and DWT techniques. the wavelet decomposition. After wavelet decomposition, many signal processing, such as compression and filter are liely to change the highfrequency wavelet coefficients. If the watermar sequence is embedded into this part, its information may be lost in the processing in sequence, which will reduce the robustness of the watermar [3]. In order to ensure the watermar has a better imperceptibility and robustness, the approximation sub-image LL3 coefficients are chosen to embed watermar. We can achieve the transform of the separable wavelet as in Figure. IV PROPOSED WATERMARKING ALGORITHM BASED ON TRIANGULAR NUMBER GENERATOR FUNCTIONS Here, the readable watermar is a qq binary image. We arrange the binary image to 0, watermar sequence wm. And the length of wm is the pxq. Original image is a mxn. ORIGINAL DATA L H LL HL III. DISCREET WAVELET TRANSFORM PRINCIPLES Wavelet transform is a time-frequency domain combined analysis method. It has multiresolution analysis features. LH HH Each level of the wavelet decomposition has four subimages with same size. Let the approximation sub-image and LH, LL stands for the HL, HH stand for the horizontal, vertical and diagonal direction high frequency detail sub-image respectively. Where the variable =,, 3, (N) is the scale or the level of Fig. () Proposed Watermaring 3 level discrete wavelet Transform V WATERMARK EMBEDDING SCHEME Consider the Hardmard matrix for the watermaring process are given below Dimensional discrete All Rights Reserved 03 IJARCET 36
ISSN: 78 33 Volume, Issue, December 03 Hadamard transform positive transform and inverse transform, such as the definition of formula () and () [5]: N NN i i i i i,, b x b u b y b v 0 () x0 y0 u v f x y i f x y u v N NN i i i i i,, b x b u b y b v 0 () x0 y0 i H(0,0) is called image bloc the DC component hardamard transform domain. Using an interactive relationship can generate higher Order transform matrix of Hadamard transform, such as the formula (3) below.,,,,3...(3) Step `A one-dimension chaotic sequence is originated from a logistic mapping u [4]. n n n The sequence has the same size as the length of the wm. Apply a threshold value, and then get 0- sequences A*. The program performs a XOR operation of this wm with the binary watermar image. X 0 and u are password. The sequence of the binary watermar image after encrypting is: Step Extracting the triangular components and the from original image. It is divided into square blocs of size 8 8 pixels. Then the HT is applied in each bloc. Then the DC value i, j, H of each bloc is collected together to get a new matrix,,...,,,...,,,,,,,............,,,,,..., Where n, m 8 8 (5) Step3 Mae the new matrix I to do a one-scale twodimension discrete wavelet transform with haar. According to quantization step value s, mae the low coefficient LL to qualified adjustment, then embed the watermar value. The detailed process is as follows: The quantified value qi, j wavelet coefficient can be obtained by: of the low-frequency LL i, j qi, j (6) s The process of embedding watermar information is as follows: If qi j w mod,, adjust the lowfrequency wavelet coefficient to LL' i, j q i, j s s (7) if mod q i, j, w adjust the low- frequency wavelet coefficient to s If LL ' i, j q i, j s 0, then LL ' i, j q i, j s s else LL ' i, j q i, j s s where i,,... m, j,,... n,,,3... p q 6 6 Step 4 Mae wavelet inverse transform. Step 5 The i, j, of each bloc can be obtained by extracting the corresponding value the wavelet inverse transform matrix, then mae HT inversetransform each sub-bloc. Changing the doubleprecision real number to unsigned 8-bit integer. Thus, obtain the color components in which watermar are embedded. Finally, we transform the image from threebasic-color image into true color RGB space. Then we will get the watermared color image. All Rights Reserved 03 IJARCET 37
ISSN: 78 33 Volume, Issue, December 03 qi j w mod,, (0) where i,,... m, j,,... n,,,3... p q 6 6 Fig. (3) Proposed Watermaring Extraction Technique VI WATERMARK EXTRACTING SCHEME: The processes of watermar extracting and embedding are reverse. When extracting watermar, the detailed ways is as follows: Step Extracting the green components (G), it is divided into 8 8 sub-bloc. Then the HT is applied in each bloc. Then the DC value, collected together to get a new matrix I ' i,,... m, j,,... n 8 8 i, j, of each bloc is,,...,,,...,,,,,,,............,,,,,..., Where n, m 8 8 Step Mae the matrix I ' to do a one-scale (8) Two-dimension discrete wavelet transforms with haar, and extracts the watermar from low-frequency wavelet coefficient LL. The detailed way is as follows: LL i, j qi, j (9) s The word s refers to quantization step value, and W'() refers to extracted watermar sequences. Step3 The watermar sequences which are extracted carry on chaotically decryption. Then it can be transformed into a binary image. Here we use the normalized correlation (NC) to measure Here we use the normalized correlation (NC) to measure the similarity between original image W and the detected watermar image W ' [6]. C n n J n n J,. i, j W i j W,., W i j W i j () In order to get rid of the impact of subjective factor, this paper adopts pea signal-to-noise ratio (PSNR) to measure the fidelity between the original image and the image which watermar is embedded. VII PEAK TO SIGNAL NOISE RATIO: Pea Signal to noise ratio (PSNR) is one of the Performance measure of the proposed method [6] and is calculated as follows PSNR 0log0 N i A,, f i j f i j All Rights Reserved 03 IJARCET 38
ISSN: 78 33 Volume, Issue, December 03 VIII SIMULATION RESULTS:.Cover image 7.MSB of cover 3 4 3 4 0 600 0 600 Fig. (4) Graphical representation of PSNR vs. Normalized coefficient From the graph we can observe a plot is made between Fig. (5) Cover Image Fig. (6) MSB of cover Image PSNR and Normalized Coeffient(NC).asPSNRincreases.WaterMar 8.Water Mared Image the NC also increases because both the linearly related Table: Comparison of PSNR values with standard value in db 0 600 700 3 4 0 600 700 800 900 0 0 600 Host image Method set jpg set tiff set 3 bmp Fig. (7) Watermar image Fig. (8) Watermared Image Proposeds 8X8 39.4 37.397 34.399 3.MSB of WaterMar 7.MSB of wmimage ystem Shereen et al. 4.5 4.6 4.7 0 600 700 3 4 0 600 700 800 900 0 0 600 H.Song et al 35.5 36.65 34.98 Fig. (9) MSB of watermar Fig. (0) MSB of Watermared image THE TABULAR COLUMNS GIVES VALUES OF PSNR OF 8*8 4.Shifted wmar 0.LSB combined with msb of wmar PIXEL OF DIFFERENT IMAGE TYPES 0 600 700 0 600 700 800 900 0 Fig.() shifted watermar Fig.() LSB combined with MSB of watermar All Rights Reserved 03 IJARCET 39
ISSN: 78 33 Volume, Issue, December 03 5.LSB of cover.lsb Extracted 3 TABLE: NORMALIZED COEFFICIENT VALUE COMPARISON BETWEEN HT-DWT-TGF AND DWT 4 0 600 Fig.(3) LSB of cover image Fig.(4) LSB extracted.predicted Watermar 3.Regenerated Cover image Host Image Method set set set 3 set 4 Bmp 6*6 Jpg 6*6 Tiff 6*6 Grey 3 4 0 600 Fig.(5) predicted watermar Fig.(6) Regenerated cover image HT-DWT- 0.9 0.88 0.7 0.83 TGF DWT 0.05 0.06 0.4 0.59 Table: PSNR Value comparison between HT-DWT- TGf Host Image Metho HT- DWT- TGF set Bmp 6*6 set Jpg 6*6 set 3 Tiff 6*6 set 4 Grey 5.5 6.7 7.6.5 DWT.0 3.0 9. 7.9 The tabular column gives PSNR values comparing between HT-DWT-TGF and DWT FOR different types of image types. The tabular column gives NC values comparing between HT-DWT-TGF and DWT FOR different types of image IX CONCLUSIONS The essential conclusion that comes from the proposed technique is the high robustness to (almost) attacs that may be implemented by media forgers. In most of the attacs, the hidden watermar could be always extracted either complete or incomplete, i.e. there is always a recognizable watermar. The proposed technique may be embedded into image, video, or audio. Another important conclusion is that, the proposed fading technique produces an exact (%) extracted watermar when rotation with (45, 90, 80, 70, 360) degrees. As a future step, the proposed technique could be embedded into video and audio watermaring. According to the experimental results and high error metrics, the novel proposed fading technique proven that it is very simple and robust against multiple attacs. Furthermore, the novel technique is the first of its type that embed a watermar that has the same dimensions All Rights Reserved 03 IJARCET 30
ISSN: 78 33 Volume, Issue, December 03 with the original cover image. It must be mentioned that the payload of the proposed technique is high, i.e. the embedded watermar does not affect the size (in Kilobytes) of the cover image. X REFERENCES [] I.J.Cox, J.Kilian, F.T.Leighton, and T.Shamoon, Secure spread spectrum watermaring for multimedia, IEEE Trans. Image Processing, vol.6, pp. 673-687, December 997. [] G.Langelaarand, R.Lagendij, Optimal differential energy watermaring of DCT encoded images and video, IEEE Trans., vol.0,pp.48-58,january. [3] J.W.Huang, Yun Q.SHI, W.D.Cheng, Image watermaring in DCT: an embedding strategy and algorithm,journal of Electronic, vol.8,pp.57-60, April 0 [4] Z.M.Zhang, L.Wang, Semiblind image watermaring algorithm in DCT domain with chaotic encryption, Computer Engineering, vol.9,pp. 9-, October 3. [5] Panaj U.Lande, Sanjay N.Talbar., FPGA Implementation of Adaptive Watermaring Using Human Visual Model,.ICGST-PDCS Journal, vol.9,pp.7-,january 9 [6] L.C.Li, Z.Y.Lu, Desynchronization attac on digital watermars and their counte measures, Journal of image and graphics, vol.0,pp.403-409, October 5. [8] F. Hartung, B. Girod, Watermaring of uncompressed and compressed video, Signal Process. 66 (3) (998) 83 30. [9] C.W. Tang, H.M. Hang, A feature-based robust digital image watermaring scheme, IEEE Trans. Signal. Process [0]C.W. Tang, H.M. Hang, A feature-based robust digital image watermaring scheme, IEEE Trans. Signal. Process. All Rights Reserved 03 IJARCET 3