ROBUST WATERMARKING TECHNIQUES FOR COLOR IMAGES

Presented By

Amit Phadikar

1

Introduction
Technologies for Security of Multimedia Data Fingerprinting Cryptography Stegnography Watermarking

2

Difference among Fingerprinting, Cryptography, Steganography and Watermarking
Fingerprinting uses some kind of hash functions to create fingerprint, original file remain intact. Cryptography is about protecting the meaning of the document. Steganography is about concealing their very existence. Watermarking is about robustness against possible attacks, Watermark need not be hidden.
3

Watermarking can be applied to
Images Text Audio S/W

Digital Image watermarking
A Digital Signal or pattern inserted into a digital image.
4

General Framework for Watermarking
Encoding Process

E(I,S)= Î

5

I)= SI 6 .Decoding Process D(J.

Comparator 7 .

Applications of image watermarking IPR Protection Demonstration of rightful ownership Authentication Labeling for data retrieval Covert communication 8 .

Properties of Digital Watermark Perceptually invisible Robustness Cost Capacity Recoverable Reversible Undetectable Able to determine the true owner High bit rate 9 .

Attacks on Digital Watermarking Lossy Compression Geometric Distortions Common Signal Processing Operations – Linear filtering such as high pass and low pass filtering – Non-linear filtering such as median filtering – Addition of a constant offset to the pixel values – Addition of Gaussian noise Jitter Attack 10 Gaussian and Non – Local exchange of pixels .

Work Already Done On This Field Watermarking domain Frequency domain Spatial domain Frequency domain Watermark is embedded in DFT. DCT and DWT domain coefficients 11 .

DCT Watermark was a sequence of 1000 random numbers.DWT Proposed to add Gaussian noise as watermark in the middle and high frequency coefficient of DWT. FM Boland:-DFT Embedded the watermark in the phase information in the discrete Fourier transform domain since the phase distortion is more sensitive to HVS than magnitude distortion . 12 . Detection was correlation based.Representative work Cox:. There fore it is more robust to tampering when compared to magnitude distortion. Watermark was embedded in the 1000 largest DCT coefficients. Correlation based non-blind detection were performed Xia Boncelet:.

Tirkel. however.bi) and increased the ai by one. certain statistical propriety like image intensity are uniformly distributed. The watermark.Spatial Domain Schyndel. The watermark was embedded to the LSB of the original image. The scheme is extremely sensitive to geometric transformation. Bender :. The watermark was detected by comparing the sum of the difference of ai and bi of the n pairs of the points with 2n provided. and Osborne :. Cross-correlation based detection was proposed.LSB Generated a watermark using an m-sequence generator. 13 . while decreased the bi by one. was not robust to additive noise.Statistical Described the patch work algorithm. it chooses randomly n pair of image point (ai.

The watermark is right-angled isosceles triangular sequence generated from a random sequence depending on a secret key.Block based Proposed partitioning both host and binary watermark Image into blocks.Feature based Proposed to find feature points and apply Delaunay tessellation to obtain the triangular sequence. setting LSB’s of each image block to zero. 14 .Spatial Domain P.Macq :. M. Detection is performed by finding Delaunay tessellation of the test image and wiener filtering to obtain watermark and then performing correlation. The watermark image block is ex-ored with the output of the hash function and output is inserted into the LSB of the image block to form watermarked image block. After applying affined transform and visual mask watermark sequence is added to the image. For extraction reverse steps are followed. J. Ping Wah Wong and Nasir Memon :.Bas. Chassery and B. Scheme is reported to detect and report any changes to the image. applying hash function (MD5) to image block.

+ ut. Furthermore.. Where vt = f(ut. u0(b) u1(0) u1(1)……….ut . the parameter B is called the encoder memory. b<= c binary convolution encoder (time-invariant and without feedback) the causal information sequence u = u0u1 ………………= u0(0) u0(1) …….B): The function f must be a linear function.. is encoded as the causal code sequence v= v0 v1 ………………= v0(0) v0(1) ……. ut – 1………….Convolution Encoding and Decoding In a general rate R = b/c. v t = ut G + ut.BG B.1G +…….. u = u0u1 ……. v1(c)………. v0(c) v1(0) v1(1)………. 0 1 15 . u1(b)…….

Probability p and the starting and ending state is predetermined to be the zero-state Why Viterbi Decoding •A highly satisfactory bit error performance. 16 .Viterbi Decoding it estimateS v1 a sequence v that maximizes P(r/ v). Where r is sequence . •Fixed decoding time. •Low cost. •Ease of implementation. •High speed of operation.

i=1.R2…….Rn. (d) P (Ri)=TRUE for i=1.2…. R1.2………n. such that n (a) U Ri=R. i=1 (c) Ri ∩Rj=Φ for all i and j.. 17 .n.Quad Tree Region Splitting Image Segmentation Method Region Based Image Segmentation Let R represent the entire image region. i≠j. then we may view region based segmentation as a process that partitions R into n sub regions. (b) Ri is a connected region.

Quad Tree Approach Quad trees Advantage of Quad Tree Decomposition Small regions represent the presence of critical information of the image and hence are the good place for the watermark insertion 18 .

Spatial Domain Watermarking watermark insertion 19 .

watermark detection 20 .

as bitij ∈ YH if I >Imean bitij ∈ YL if I ≤ Imea whereYH and YL are the high and low intensity classes. Step 2: Classify each pixel into one of two categories.Watermark Insertion Algorithm Apply QUAD TREE decomposition on color image I (x. I. bitij is classified depending on its intensity. the ijth pixel. based on whether its intensity value is above or below the mean intensity of the block. Imean. Imin. of the the pixels in H.e. minimum. i. Imax. respectively. and maximum. 21 . Watermark embedding region Repeat (for each selected 4x4 block (H) of blue channel) { Step 1: Compute the average. y) and select all 4x4 blocks in blue channels..

λ if I < meanL I new = Imean . Step 7: The modified block of pixels. YL and YH. of pixels from the original host image. is then positioned the watermark image in the same location as the block.λ if Imean ≤I < meanH I new = I .Step3: Compute the means. if I > meanH I new = Imax + λ I new =Imean + λ if meanL ≤ I < Imean I new =I + δ otherwise if bitw = 0. Step 6: Given the value of bitw is 0 or 1. modify the pixels in H according to: if bitw = 1. meanL and meanH. Hnew. 22 Step 8: Marge red. Step 4: Define the contrast value of block H as CB = max(Cmin. β(Imax-Imin)) where β is a constant and Cmin is a constant which defines the minimal value a pixel's Intensity can be modified. for the two classes. green and blue channel.δ otherwise Where I new is the new intensity value for the pixel which had original intensity value I and δ is a random value between 0 and CB and λ is the watermark strength. } Until all watermark bits are inserted. Step5: Select a watermark bit (bitw ) randomly depending on the key value. H. I new = Imin . .

Y) values lies in the range Xmin +(Xmax. maximum coordinate value in X and Yaxis of that image. Step2: a watermark bit is decoded by making the comparison of the two resultant values: If Average w > Averageo. Ymax are the minimum . The decoded bits are then arranged in order using same key. Repeat{ Step 1: take one 4x4 blocks of the host image and Corresponding 4x4 block of watermarked Image using the same coordinate value as of 4x4 block of host image . Xmax.(Ymax. Ymin. and whose all pixel coordinate (X. then bit w = 0 Where Average o and Average w are the averages for the 4x4 blocks of the host and Corresponding 4x4 block of watermarked Images. which was used during embedding. } Until all watermark bit are extracted.Ymin )/4 <=Y<=Ymax .Ymin )/4 where Xmin. respectively. y) and select all 4x4 blocks in blue channels that passes the homogeneity test.Xmin )/4 and Ymin +(Ymax.Xmin )/4 <=X<=Xmax (Xmax. then bit w = 1 If Average w ≤ Average o. Then. 23 .Watermark Extraction Algorithm Apply QUAD TREE decomposition on original image I (x. the encoded watermark is exclusive ored by 128 bit key and then decoded by viterbi decoding.

β=1 and λ=15 and convolution encoding rate R=1/2. Test Image LENA.BMP (512X512) .Watermark is a 50x50 binary bitmap NCC =∑∑ W ij ij W' ij / ∑∑Wij ] ij 2 (a) (b) (c) (d) Fig (a) original or host image (b) watermark image (c) watermarked images (d) Extracted watermark 24 .Results: Cmin =15.

mask(3x3) Median filtered watermarked image and extracted watermark.PSNR VS λ 45 44 43 42 41 40 0 5 10 λ 15 20 (e) PSNR (f) Fig (e) watermark embedded region. Wiener filtered watermarked image and extracted watermark .t.mask(3x3) 25 . Various values of λ. (f) the variation of PSNR w.r.

75 Cropped watermarked image and extracted watermark .Scaled down watermarked image. mask(444x444) Jpeg compressed watermarked image and extracted watermark 26 . scale down factor=. rescaled image and extracted watermark.

angle=-17 degree Table Showing the Normalized Cross correlation values for different operations Table lists the PSNR between original host image and watermarked image for various value of λ 27 . rotation corrected image and extracted watermark.Rotated watermarked image.

Graphs for different operations showing variation of NCC value against various factors 28 .

the following orthogonality condition on the filters and 29 .Wavelet Domain Watermarking Discrete wavelet Transform (DWT) The DWT and IDWT can be mathematically stated as follows A signal. x [n] can be decomposed recursively as and the signal x [n] can be reconstructed from its DWT coefficients recursively To ensure the IDWT and DWT relationship.

DWT pyramid decomposition of an image. 30 .Figure.

Figure. Examples of a DWT pyramid decomposition 31 .

watermark insertion 32 .

watermark detection 33 .

T) T=20 is the threshold value used by Quadtree and bi denote ith block such that 1=<i<=M and function Quadtree is used for quad tree decomposition of image in spatial domain. 34 . y). b2…. which is used to embed a watermark is segmented by quad tree decomposition to select all 4x4 blocks Watermark embedding region B'= [b1..Watermark Insertion Algorithm The host image B (x.bM ] = Quadtree (B.

green and blue channel to get the watermarked image 35 . c'h = ch . 2: compute average(C avg ) of all coefficient Ci found in step 1. 5: Given the value of bitw is 0 or 1. C avg = 1/16 Σ Ci 3: for all coefficients CH Є Ci such that CH > C avg 4: Select a watermark bit (bitw ) randomly depending on the key (k2) value from watermark bit sequence (W''' 2N ). 7: The original block of pixels bi is then replaced by b'i } Until all watermark bits are inserted. c'h = ch + λ* ch if bitw = 0. Repeat (for blue component of each selected 4x4 block bi Є B') {1: perform single scale wavelet transform of block bi .The bit embedding strategy is as follows. modify all the coefficients ch Є CH according to: if bitw = 1.λ* ch Where c'h is the new value of coefficient in CH which had original Coefficient value of ch and λ is the watermark strength 6: perform inverse single scale wavelet transform after modification of Coefficient to get modified block b'i. 8: Marge red.

4: initialize variable sum_w=0 and sum_o =0. Xmax. where WT denotes single level wavelet transform. Y) lies in the range Xmin +(Xmax.Xmin )/4 and Ymin +(Ymax. 5: for j=1:16 if Ci (j) > C avg sum_w = sum_w + C'i (j).(Xmax. 2: perform single level wavelet transform of block bi and b'i to get the Coefficient in wavelet domain which is defined as Ci = WT(bi) C'i = WT(b'i). Ymax are the minimum . end end 36 . and whose all pixels (X. y) and select all 4x4 blocks that passes the homogeneity test. sum_o = sum_o + Ci (j).Ymin )/4 where Xmin.Ymin )/4 <= Y <=Ymax .Watermark Extraction Algorithm Apply quad tree decomposition on original image B (x.(Ymax. Ymin. Ci and C'i are the vectors of the same length representing wavelet coefficient 3: compute average of all coefficient C avg of Ci .Xmin )/4 <= X <= Xmax . maximum coordinate value in X and Y axis of the image Repeat { 1: Take one 4x4 blocks (bi) of the host image and Corresponding 4x4 block (b'i) of watermarked Image using the same coordinate value as of 4x4 block of host image .

Then. } Until all watermark bit are extracted. 8: W2 [pos]= bit w .6: a watermark bit ( bit w) is decoded by making the comparison if sum_w < sum_o. bit w = 1 7: find original position (pos) of extracted watermark bit ( bitw ) using Key (k2) as seed. Where W2 is an array for storing extracted Watermark bits. the encoded watermark is exclusive ored by 128 bit key and then decoded by viterbi decoding. then bit w = 0 if sum_w >= sum_o. 37 .

2 λ 0.Results: λ=.tif (512X512) . Test Image kids.4 (d) 0. (f) the variation of PSNR w. various values of λ.t.Watermark is a 50x50 binary bitmap (a) (b) (c) PSNR VS λ 60 PSNR 40 20 0 0 0.6 (e) (f) Fig (a) original or host image (b) watermark image (c) watermarked images (d) Extracted watermark (e) watermark embedded region.3 and convolution encoding rate R=1/2. 38 .r.

rescaled image and extracted watermark.75 39 .scale down factor=.Wiener filtered watermarked image and extracted watermark.mask(3x3) Scaled down watermarked image.mask(3x3) Median filtered watermarked image and extracted watermark.

mask(444x444) Jpeg compressed watermarked image and extracted watermark 40 .Cropped watermarked image and extracted watermark .

Rotated watermarked image. watermark.angle=-10 degree rotation corrected image and extracted Table Showing the Normalized Cross correlation values for different operations Table lists the PSNR between original host image and watermarked image for various value of λ 41 .

5 0 0 50 100 Angle in Degree Lossy JPEG Com pression 1.5 0 0 5 Mask Size(N XN) 10 0.95 0.5 0 0 1 2 3 4 Scale Up Factor Rotation in Positive Direction 1.5 1 0.5 0 0 50 100 Angle in Degree NCC NCC Fig Graphs for different operations showing variation of NCC value against various factors 42 NCC NCC .Wiener Filter 1.5 Scale Dow n Factor 1 Scale Up 1.5 NCC NCC 1.9 0.9 0 Scale Dow n NCC 1 0.5 0 0 5 Mask Size(N X N) 10 Median Filter 1.5 1 0.7 0.05 1 0.5 0 50 JPEG 100 150 Rotation in Negative Direction 1.5 1 0.1 0.5 1 0.

few algorithms resist the watermark copy attack and ambiguity attack. Bulk of the literature contains linear additive watermarks. 43 . image But none of the techniques proposed so far seems to be robust to all possible attacks Most of the current methods require the original image at watermark recovery this may prove as serious limitation at certain moments. as shown by the results. as the image may not be accessible Very few algorithms deal with the problems associated with geometric changes. Consequently further work needs to be done to improve the robustness of algorithms. Very few algorithms proposed in the literature consistently survive the random bending attacks.Conclusion and Research Directions Both the algorithms are robust to common processing operations.

. New York.p-df [2] Daniel Gruhl. T. September 2000. http://www. In Anderson [66].. Northwood.J.edu/~smohanty/research/Reports/WMSurvey1999Mohant y.L. pages 295-315.vu. 1999.html [4] M. 1999. Dec. 1996. M. www. Information Hiding. Technical report. Petitcolas and Markus G.K. MA: Artec House. Parhi.cs. S. K. Kuhn. Echo hiding.A. “Introduction to Watermarking Techniques” in Information Techniques for Steganography and Digital Watermarking. pp 58-64. http://www. "Digital Watermarking: A Tutorial Review". P.ac.usf. Marcel Dekker. Kutter. eds.edu /~shoemakc /watermarking / watermarking. pp 97-119 [5] I. Inc. 44 [1] . Second International Workshop.uct.csee. 1998.P. Hartung. pp 461-482 [6] F. Miller. Kalker. “A Review of Watermarking Principles and Practices” in Digital Signal Processing for Multimedia Systems.html [8] Ross J. “Attacks on copyright marking systems”. F. [7] Chris Shoemaker “Hidden Bits: A Survey of Techniques for Digital Watermarking”. [3] James Greenfield “Digital Watermarking”.. and Anthony Lu. T.G. Walter Bender. Technical report.za/courses/CS400W/NIS/papers99/jgreenfi/watermarking. Anderson Fabien A. New York. Vol 17. Linnartz. Petitcolas. Cox.References Saraju P.C. in IEEE Signal Processing Magazine. Mohanty.M. Eds. “Watermarking Schemes Evaluation” ”.union. Katzenbeisser et al. J. Nishitani.

vol. on Image Processing. Cox. Cox. “Using the Fractal Code to Watermark Images”. 1999. Oct. and T. no. 1673-1687. I. and B. “Evaluation of copyright marking systems”. Jul. Boncelet and G. Shamoon. 12. Davoine. “Psychovisual Approach to DigitalPicture Watermarking. 1997. McKellips. [14] I. [10] Bhupendra Verma “Implementation of Digital image watermarking techniques”. Mitrea. M. pp. Miller and A. 469-473. pattern recognition and application 45 (SPPRA’02) Crete Greece. [13] X. Proc. 1998. “Phase watermarking of digital images”. Kilian. vol.P. “A Multiresolution Watermark for Digital Images”. I. 1127-1141. Arce. no. Preteux. C. Conf. [15] I J. 3. Conf. Vol 3. Presented at IEEE Multimedia Systems. Macq.). “Watermarking as communications with side information”. C. 7.References [9] Fabien A. Bas. 1997. J. 548-551. [11] P. J. vol.” Journal of Electronic Imaging. and F. JULY 2003. 2002. on Image Processing. 1999. P. pp. Mtech thesis submitted to RGPV (Technical university of M. F. IEEE Int. “Secure Spread Spectrum Watermarking for Multimedia. . IEEE Int. 239-242. “Spread Spectrum Watermarking method for image databases”. 6. pp. F.J.P. Proceedings IASTED international conference on signal processing. Proc. L. O’Ruanaidh. L.Dowling and FM. Boland . Chassery. [12] J. Leighton. July 1998. Rougon. pp. vol. Proceedings of the IEEE. pp. [16] JJK. pp. De Vleeschouwer. Anderson.” IEEE Transactions on Image Processing.1996. Dec. Vlad and N.in Proceeding of ICIP’96. Delaigle. [17] M. A. Oct. pp 444-449. 628-640. W J. Xia. Petitcolas and Ross J.

Mitrea. New Jersey. [23] C. pp 311-318 [20] X.References [18] C. Zeng. Oct.323. [22] M. Jun. [21] F. 450-454. “ Multibit digital watermarking robust against local nonlinier geometrical distortions”. 1998. IEEE Int. pp 999-1002. “A Multiresolution Watermark for Digital Images.Pun. on Image Processing. 1. Ebrahimi. “DCT based watermarking for video”. 1997. Boncelet. 2002. 1998. Conf.Kutter. The First IEEE Signal Processing Society Workshop on Multimedia Signal Processing. Proc. 44. K. in IEEE ICIP 2001. pp 320. Cappellini and A. Vlad. vol. Greece. Wu. I. “ Robust Watermarking Method for Color still image database”. pp 206-216. on Image Processing. 1997. C. S. Int.I.P. Bartolini. Conf. Podilchuk and W.T. On Image Processing. Piva. Oct. F. Voloshynovskiy F. Proc. Hsu and J. Proceedings 2002 tyrrhenian International Workshop on Digital Communications advanced methods: for multimedia signal Processing ( IWDC’02) Capri Italy . pp. [24] S. 548-551. Bhattachrjee and T. M. vol. vol. “ Toward second Generation Watermarking Scheme”. Xia.” Proc. Deguillaume and T. Thessaloniki. Princeton. 46 . Preteux and A. ”Perceptual watermarking of still images”. Proc. IEEE Int.No. Vol. and G. 1999. Arce. V. pp. oct 2001. “Mask Building for Perceptually Hiding Frequency Embedded Watermarks”. Conf. IEEE Transactions on consumer Electronics. I. I. Barni. Feb. [19] M.

J. Lin. San Jose.J. 3022. Zhao and A. 1999. Mckellips.Lui. U.Bloom. vol 87. No. Saddik. [30] Ching -Tang Hsieh and Yeh-Kuang Wu “ Digital Image Multiresolution Watermark Based on Human Visual System Using Error Correcting Code” Tamkang Journal of Science and Engineering. 7. [26] F. no 5. “ Denoising and copy attacks resilient watermarking by Exploiting Knowledge at Detector”.Cox.L. “ RST Invariant Digital Image Watermarking Based on Log-Polar Mapping and Phase Correction”. F.. Quisquater. B.A. Guelvouit and Stephane Pateux. Goffin. “ Rotation. [27] Dong Zheng.J. Miller and A. IEEE transactions on image processing. M. Feb. Mark Liao and M. Y. 2001. of SPIE Santa clara CA. vol 10.L. The Society for Imaging Science and Technology (IS&T) and the International Society for Optical Engineering. De Vleeschouwer. of IEEE. pp.References [25] C. 3. Wu. 1997. 3. M. “ Wide spread spectrum watermarking with side information and interference cancellation”. Scale. “A low cost perceptive digital picture watermarking method”.Miler and Y. Macq and J. no. 201-208 (2001) [31] I. pp.M. 11. Proc. IEEE Trans.Kutter.S. no. [28] Chun Shien Lu. I. J. California. C. J. September 20003. Proc. J. Y. 47 . E. IEEE Transc. L. 4. “ Watermarking as communications with side information”. no. H. Jan 2003. pp 767-782.A. vol XX. pp 1127-1141. 264-277.Cox. vol. Y. Vol. Delaigle. On circuits and systems for video technology. March 2002 pp 280-292 [29] G.L. M. and Translation Resilient Watermarking for Images”. On Image Processing. vol.

35(3). “Spacial method for copyright labeling of digital images”.Kutter. Quisquater and B.326332.456-459. “A watermark for digital images”.Lausanne. In Dr.J. 1996. and A. In Proceeding of the IEEE Int.Mormoto.Lu. “Image authentication for a slippery new age”.Conf. Bender.June 1995. “Low cost spatial watermarking”. Davern and M. Scott. No 2. in Proceedings of ICIP’96.B. Journal of Electronic Imaging.References [32] M. J. 1998.No 4.OnImageProcessingICIP-96. IBM Systems Journal . N. [35] I.Macq and J.Delaigle.Walton. pp.215-218. Dobb’s Jurnal. Quisquater.J. Computer and Graphics.”Digital watermarking of color images using amplitude modulation”.Delp. Darmstaedter.Vol.Switzerland. 3.417-424.J.F. [38] P. 7. Jordan and F.M.1996. 48 .pages21518.pages 944-957. “Techniques for data hiding”. [33] S. F.Wolfgang and E.Bossen. Vol. “Fractal based image steganography”. and B. [36] R. [40] V.in Proceeding of First International Workshop on Information Hiding .Macq.Bruyndonckx.volume 83. in Proceedings of IEEE Nonlinear Signal Processing Workshop.pp. [37] W. Vol.J. pp 16-19 September 1996.22. pages 18-26 April 1995.pp.1998. [34] B. D. Patas “A method for signature casting on digital images”.1997. pp 313-336.pp 273-294. [39] O.J. In Proceeding of the IEEE. Gruhl. J. Quisquater. pp. Macq. “Cryptology for digital tv broadcasting”.1995.

2002. Osborne. “ Secret and public key image watermarking schemes for image Authentication and ownership verification” IEEE transactions on image processing. J. IS&T/SPIE Proceedings. 10 no. 86-90. 1994. [42] G. Nov. Feb. Enschede. Chassery and B. CA. S. C. van der Lubbe and J. IEEE int. on Image Processing. “ Geometrically Invariant Watermarking Using Feature Points”. C. In Proc. Information Theory in Benelux. 1999. L. Conf. Voloshynovskiy and T. [44] R.Macq.Bas. [47] P. [46] F. A. Lagendijk. Biemond. II. 11. pp. In IEEE Transactions on Consumer Electronic. J. [43] C. October 2001. of 7th Symposium. A. USA. pp 1593-1600. “Robust labeling methods for copy protection of images in Storage and Retrieval for Image and Video Databases”. Langelaar. C. France 2002. “Copy protection for multimedia data based on labeling techniques”. vol. San Jose. pp.References [41] G. J. vol. 298-309. Lee and Y. pp 101449 1028. 9. van der Lubbe and R. “A Digital Watermark. Tirkel. Schyndel. The Netherlands. “An Adaptive Digital Image Watermarking Technique for Copyright Protection”. 10. and C. C. . Proc. May. 4. Of European Signal Processing Conf. On Image Processing.No. Deguillaume. No. “ Hybrid robust watermarking resistant against copy attack”. Nov. 1997. A. M. vol. 45.K.Pun. [45] Ping Wah Wong and Nasir Memon. Vol. Langelaar. Lee.H. IEEE Trans. 1996.” Proc.

2005. Woods and Steven L. Gonzalez. Johansson and F. F. [55] C. 1996. Johannesson.References [48] O. pp. 326-330. K. “ Hidden signatures in images”. pp 223. [50] Ranjan Bose. In Proc.T. Jönsson. Zigangirov.f/viterbi/tutorial. SPIE Int. Eddins “Digital Image Proceing Using Matlab” Pearson Education. Boland and C. rotation and scaling”. Symposium on voice. 2002. in proceedings of ICIP’96. New York. 1999.Hsu and J. [54] Rafael C. “Fundamentals of Convolutional Codes”. Springer-Verlag.I. EUROCRYPT'99. [53] T. Proceedings of the IEE Conference on Image Processing and its Applications.. Nov 1998. Video and Data Communication. "Improved fast correlation attacks on stream ciphers via convolution codes".226.html [52] R.netcom.Tata McGraw-Hill. 50 . Richard E. Edinburgh. Information Theory Coding and Cryptography.Wu. 1995. Ruanaidh J. “Watermarking resistant to translation. [51] Chip Fleming “A Tutorial on Convolutional Coding with Viterbi Decoding” http://home. "Watermarking Digital Images for Copyright Protection". Sh. IEEE Press.com/~chip. 1999. Kutter. Dautzenberg. LNCS 1592. [49] M.

315. Vaidyanathan. vol. “Multirate Systems and Filter Banks”. Shapiro. pp 69-87. pp 14-38. “Wavelets and Filter Banks”.References [56] Saenz. Prentice Hall. Charles G. 1995. Cambridge. pp. on Pure and Appl. [58] I. In: Gabbouj. No. Philadelphia. 1989. Vol. Vetterli. Boncelet and Gonzalo R. “Color image wavelet compression using vector morphology”. Daubechies. Wellesley-Cambridge Press. 12. Daubechies. 1996. J. Soc. [60] I. “Embedded image coding using zerotrees of wavelet coefficients.bridge. “Wavelets and signal processing.. [65] Xiang-Gen Xia. Kovacevic. Math. Nguyen. 1174 of 51 lecture notes in computer science. Öktem. M. December 1998 [66] Ross J. (eds). R. 1991. E. Englewood Cliffs. On Signal Processing. Vetterli and J. Prentice Hall. M. [62] M. Ten Lectures on Wavelets. OPTICS EXPRESS”. Q. 1996. Anderson. 1993. 3. Math. Strang and T. [57] S. Rioul and M. 4-8 September 2000. “Orthonormal bases of compactly supported wavelets. & Delp.. Cam. 41. . & Kuosmanen. Englewood Cliffs. Mallat. [61] P. P. 41." IEEE Signal Processing Magazine. Technical report. (SIAM. Signal Processing X Theories and Applications. K." IEEE Trans. [59] O. NJ. “Wavelets and Subband Coding”. Finlands. Arce “Wavelet transform based watermark for digital images.. pp 3445-3462 1993. pp 1988.. “Multiresolution approximations and wavelet orthonormal bases of L2(R)." Comm. 1992). [64] J. NJ. 909-996.tenth European Signal Processing Conference. Proceedings of EUSIPCO 2000. “Information hiding” first international workshop. Tampere. Isaac Newton Institute. 115-118. Egiazarian. [63] G." Trans. P. Amer.

2. 1. Co-author(s): Bhupendra Verma. Sanjeev Jain 52 . accepted for publication. ISSN 0973-3663. accepted in International Conference on Information Security (ICIS). st is tanbul. Sanjeev Jain 2) A New Color Image Watermarking Scheme. 2006. an International Journal). pp 45-50.Publication Paper accepted 1) Spatial Domain Robust Image Watermarking Scheme. Co-author(s): Bhupendra Verma. ADIT Journal of Engineering. Dec 2005. 24-26 June Co-author(s): Bhupendra Verma. Sanjeev Jain 3) A spatial domain non-oblivious robust image-watermarking scheme.no. vol. Turkey. INFOCOMP Journal of Computer Science (INFOCOMP.

Sanjeev Jain 3) Region Splitting Approach to Robust Color Image Watermarking In Wavelet Domain. Co-author(s): Bhupendra Verma. Sanjeev Jain 2) Quad Tree Region Splitting Approach to Spatial Domain Robust Color Image Watermarking. International Journal of Information Technology. International Journal of Image and Graphics (IJIG). Publisher Elsevier) Co-author(s): Bhupendra Verma. (Publisher Computer Society of Singapore). Sanjeev Jain 53 . Co-author(s): Bhupendra Verma. (Publisher World Scientific).Papers Communicated 1) Spatial domain robust blind Watermarking scheme for color image. Image and Vision Computing (International Journal.

THANKS 54 .