With the result of advancement in today’s technology, digital content can be easily copied, modified, or distributed. Digital watermarking provides the solution to this problem. Most of the digital watermarking methods are divided into two categories: Robust watermarking and fragile watermarking. As a special subset of fragile watermarking, reversible watermarking (lossless or invertible watermarking) enables us to recover the image which is same as the original image pixel by pixel after the content is authenticated. This type of lossless recovery is compulsory in sensitive imagery applications like medical and military purposes.
An efficient watermarking algorithm has been implemented using Matlab which uses the concept of difference expansion of high pass transform coefficients with watermark bits. This work was to find a reversible watermarking algorithm for JPEG2000 standard for medical applications, a (5,3) wavelet transform is used which is considered as lossless transform in the JPEG2000 standard. In the algorithm, (5,3) Integer wavelet transformed high pass coefficients are difference expanded instead of Haar wavelet transformed coefficients in Mark Tian’s algorithm. Based on the Algorithm developed for Matlab modeling, a new architecture for Reversible watermarking was designed and the hardware modeling for that architecture was done using Verilog HDL. By difference expanding the high pass coefficients of the image and embedding the watermark in those high pass coefficient, maximum embedding capacity over 90000 bits is achieved for a 256x256 image.
The Watermark embedding block is synthesized using Xilinx ISE and implemented on Spartan3 FPGA. The Reversible Watermarking Block operates at a maximum clock frequency of 62.073 MHz with a minimum period of 16.110ns. The Latency of the system is N+2 clock cycles for a total of N pixels macro block. The embedding capacity of 2bits at a time are used to embed in the high frequency coefficients, as number of bits to be embedded increases the Peak Signal to Noise Ratio decreases up to 31.3%.
Watermarking
Since the research field is still relatively young and has contributors from several disciplines with varying traditions, the terminology used is still quite diverse. This section provides a formal introduction to watermarking and their types and classification.
Watermarking – The term stands for the hiding of some message or other kind of information in to an image. This information is called as Watermarking and the image that hosts it is the Cover image.
Types of watermarks
- Robust watermarks are designed to resist against heterogeneous manipulations, all applications presupposing security of the watermarking systems require this type of watermark.
- Fragile watermarks are embedded with very low robustness. Therefore, this type of watermark can be destroyed even by the slightest manipulations. In this sense they are comparable to the hidden messages in steganographic methods. They can be used to check the integrity of objects.
- Public and private watermarks are differentiated in accordance with the secrecy requirements for the key used to embed and retrieve markings.
According to the basic principle of watermarking, the same key is used in the encoding and decoding process. If the key is known, this type of watermark is referred to as public, and if the key is hidden, as private watermarks. Public watermarks can be used in applications that do not have security-relevant requirements.
- Visible or localized watermarks can be logos or overlay images in the field of image or video watermarking. Due to the implicit localization of the information, these watermarks are not robust.
1.3 Classification of Watermarking
According to signal processing methods [2]
1) Spatial domain – This technique is susceptible to image processing operations like image compression, cropping, filtering.
2) Transform domain – The mark embedding in frequency domain like DCT and DWT.
According to application point of view [3]
1) Robust Watermarking – Is mainly aimed at Copyright protection.
2) Fragile Watermarking – Is mainly aimed at content authentication. It can be altered or destroyed when the digital content is modified.
Besides the various types and classifications of watermarks, four different watermarking systems are classified according to the input and output during the detection process. Using more information at the detector site increases the reliability of the whole watermarking system but limits the practicability of the watermarking approach on the embedder side. Figure 1.2 Shows the Watermark Decoder.
The side information in the detection process can be the original co and the watermark w itself (see Figure below). Therefore, four permutations of side information requirements are possible.
4 Applications of Watermarking
Watermarking techniques are applied to images because of various reasons. Each of these possible applications involves typical processing operations that a watermarking technique must survive. Content protection scenarios may include operations like color to gray-scale conversion, global or local affine transforms, and printing and scanning. Authentication watermarks must not be affected by legal operations, while illegal attacks must destroy them. Metadata labeling scenarios may include media transform. A typical example is the transmission of information in printed images. This information is revealed if the printed image is shown to a webcam whose data is processed with the watermark reader software as presented by Digi-marc [5].
Yet robustness is not a general requirement for data hiding techniques: Undetectibility is essential. A typical scenario for data hiding is the distribution of hidden information via newsgroups, bulletin boards, or simply by images on homepages. Steganalysis is a new research area dealing with the detection of hidden data as presented, for example, by Fridrich and Goljan [6]. A possible application of these techniques is the so-called StegoWall as proposed by Voloshynovskiy et al. [7]. This StegoWall can be compared with a firewall that analyzes the data that should be transmitted and prevents the transmission of any data containing hidden information.
1.5 Reversible Watermarking
Reversible watermarking is a subset of fragile watermarking. It has an additional advantage of recovering the image which is same as the original image pixel by pixel, after the image is authenticated.
Merits of Reversible Watermarking
- Distortion free data embedding.
- Application point of view, it can be used as an information carrier.
1.5.1 Reversible watermarking process
Watermarking for valuable and sensitive images such as military and medical image presents a major challenge to most of watermarking algorithms. First, such applications may require the embedding of several kilobytes of data, but most of robust watermarking algorithms can embed only several hundred bits of data. Second, the watermarking process usually introduces a slight but irreversible degradation in the original image. This degradation may cause the loss of significant artifacts in military and medical images. These artifacts may be crucial for an accurate diagnosis from the medical images or for an accurate analysis of the military images.
As a basic requirement, the quality degradation on the digital content after data embedding should be low. A feature of reversible data embedding is the reversibility, that is, when the digital content has been authenticated, one can remove the embedded data to restore the original content.
The motivation of reversible data embedding is distortion-free data embedding. In sensitive images such as military and medical image, every bit of information is important. Reversible data embedding will provide the original data when the digital content is authenticated.
6 Organization
The objective of the project is to insert a watermarking block in the JPEG2000 coding pipeline to reversibly embed the watermark in the digital image. This method employs a discrete wavelet transform to remove redundancy in a digital image to allocate space for watermark embedding. The watermark could be a signature of the image or sequence of bits.
2.1 Review of existing methods
The earliest reference to reversible data embedding we could find is the Barton patent, filed in 1994. In his invention, the bits to be overlayed will be compressed and added to the bit string, which will be embedded into the data block. Honsinger, et al., reconstruct the payload from an embedded image, and then subtract the payload from the embedded image to losslessly recover the original image. Macq proposes an extension to the patchwork algorithm to achieve reversible data embedding. Fridrich, et al., develop a high capacity reversible data-embedding technique based on embedding message on bits in the status of group of pixels. They also describe two reversible data-embedding techniques for lossy image format JPEG. Kalker, et al., provide some theoretical capacity limits of lossless data compression based reversible data embedding [5] and give a practical code construction. Celik, et al., present a high capacity, low distortion reversible data-embedding algorithm by compressing quantization residues.
In the past few years different reversible watermarking techniques [3] had been proposed. But all of these used to remove bits from block of the image and by losslessly compressing these bits providing the space for the watermark to be embedded in the same block. Tian proposed an algorithm [9], [16] using difference expansion and using Haar wavelet transform. This method chooses the expandable coefficients and embeds an extra bit into these coefficients. The watermarked image formed in this method is imperceptible and exact recovery is also possible. But the embedding capacity in this method is less. Alattar [17] proposed an algorithm which uses difference expansion of a generalized integer wavelet transform there by embedding a set of watermark bits in a vector of pixels.
A new algorithm is suggested for reversible watermarking which makes use of difference expansion concept in Tian’s algorithm and the concept of watermark embedding in a set of pixels presented in Alattar’s algorithm. The watermarking embedding and recovery blocks can be introduced in JPEG2000 standard without making many changes to the normal coding flow. In the suggested algorithm, the watermark bits are embedded in the transformed coefficients which don’t cause any overflow or underflow [9]. Table 2.1 the list of papers that has been reviewed.
2.2 Current Status and Key Issues
In recent years a special kind of digital watermarking is discussed widely, called reversible watermarking. It not only provides the protection of the copyright by embedding the assigned watermark into the original image but also can recover the original image from the suspected image. The retrieved watermark can be used to determine the ownership by comparing the retrieved watermark with the assigned one. Similar to conventional watermarking schemes, reversible watermarking schemes have to be robust against the intentional or the unintentional attacks, and should be imperceptible to avoid the attraction of attacks and value lost. Therefore, the reversible watermarking also has to satisfy all requirements of the conventional watermarking such as robustness, imperceptibility, and readily embedding and retrieving. [12]
Except for these requirements, reversible watermarking has to gratify the following two additional requirements.
1) Blind:
Some of the conventional watermarking schemes require the help of an original image to retrieve the embedded watermark. However, the reversible watermarking can recover the original image from the watermarked image directly. Therefore, the reversible watermarking is blind, which means the retrieval process does not need the original image.
2) Higher Embedding Capacity:
The capable size of embedding information is defined as the embedding capacity. Due to the reversible watermarking schemes having to embed the recovery information and watermark information into the original image, the required embedding capacity of the reversible watermarking schemes is much more than the conventional watermarking schemes. The embedding capacity should not be extremely low to affect the accuracy of the retrieved watermark and the recovered image.
2.3 Discussions [12]
The higher embedding capacity usually comes along with a higher distortion. Therefore, based on the watermarked images with the similar image quality, we compare the embedding capacity with the three types
4 Research Issues [12]
It is clear from the above discussion that the embedding capacity and the robustness are the two major challenges of reversible watermarking. Furthermore, many of the conventional watermarking schemes are embedding watermarks in frequency domains. In general, the advantages of embedding watermarks in frequency domains are naturally resisting some attacks, immune to several destructions,
Summary of Literature Review
- The PSNR (Peaks of the Signal-to-Noise Ratio) is a popular index term to valuate the difference between the pre-processing image and the post-processing image. A larger value of PSNR means that the watermarked image has a better quality, the difference between the original image and the watermarked image is imperceptible.
- The embedding capacity is 27% (2.17 bpp) for
- The advantages of embedding watermarks in frequency domains are naturally resisting some attacks, immune to several destructions, or others.
The advantage of the DWT over Fourier transformation is that it performs multi-resolution analysis of signals with localization both in time and frequency, popularly known as time-frequency localization
3.1 Introduction
Reversible watermarking is a special subset of fragile watermarking. Like all fragile watermarks, it can be used for digital content authentication. But it has an additional advantage, after content authentication one can remove the watermark to retrieve the image which is exactly same as the original image.
The motivation of reversible data embedding is distortion-free data embedding. Though imperceptible, embedding some data will inevitably change the original content. Even a very slight change in pixel values may not be desirable, especially in sensitive imagery, such as military data and medical data. In such a scenario, every bit of information is important. Any change will affect the intelligence of the image, and the access to the original, raw data is always required.
3.2 Problem statement
“Design and Implementation of Reversible Watermarking for JPEG-2000 Compression Standard on FPGA”
3.3 Objectives
• To review the literature on DWT and reversible watermarking for JPEG 2000 Standard
• To arrive at design specifications for reversible watermarking for JPEG-2000 compression standard
• To identify the suitable architecture for reversible watermarking technique
• To develop and simulate software reference model for the identified reversible watermarking architecture for the JPEG-2000 compression standard
• To model the reversible watermarking architecture and simulate the hardware module of the design developed for the JPEG-2000 compression standard keeping software reference model as reference.
• To implement the design in FPGA and verify the performance requirements
3.4 Methodology
• Literature on various algorithms and techniques of reversible watermarking for JPEG-2000 compression standard reviewed by referring to books, Journals, conference papers, websites
• Suitable algorithm for reversible watermarking of JPEG-2000 compression standard has been identified based on reviewed literature
• Design specifications for reversible watermarking has been derived based on reviewed literature and JPEG-2000 compression standard
• Architecture for reversible watermarking of JPEG-2000 compressed image was designed based on design specifications
• Sub-modules of reversible watermarking architecture were identified
• Software reference model for sub-modules of reversible watermarking was developed from the identified architecture and algorithm using Matlab
• Sub-modules of the identified architecture of reversible watermarking were modeled using Verilog HDL
• Functionality of the developed modules was verified by simulating them using ModelSim
• Sub-modules were integrated to design the top module of the reversible watermarking and test bench was developed in verilog for its verification
• Top module of reversible watermarking was simulated and the results were verified against the specifications
• A test environment for verifying the reversible watermarking on FPGA was created
• Verilog HDL of reversible watermarking for JPEG-2000 compression standard was implemented on Xilinx FPGA
• Reversible watermarking for JPEG-2000 compression standard implemented on Xilinx FPGA was verified against design specifications and the Verilog HDL simulation results
4 – Design of Software Reference Model (Matlab Modelling)
Tian’s Algorithm - The concept of difference expansion is introduced by Mark Tian [6],[7] in his watermarking algorithm which uses Haar Wavelet transform to embed watermark bits was used as a reference algorithm. In his algorithm, a pair of pixels are used to form low pass coefficient(l) and high pass coefficient(h). The high pass coefficient(h) is difference expanded by watermark bit(b) as
h’ = 2 * h + b, where h’ is watermarked coefficient.
These watermarked high pass coefficients and original low pass coefficients are used to form the watermarked pixels, if there is no overflow or underflow that is the pixel value lies between 0 and 255 for a gray scale image, in these pixel values then the coefficient h is expandable and a watermark can be embedded in this coefficient.
4.1 Implementation
In the watermarking embedding stage, 1D (5, 3) DWT of the input image is calculated and the watermark is embedded in the high pass coefficients to give the watermark coefficients. The other coding operations are performed on these coefficients to generate coded image.
4.1.1 Embedding procedure
Algorithm Steps
i. Find the (5, 3) wavelet transform of the original image, to form low frequency (Y (2n)) and high frequency (Y (2n+1)) coefficients.
ii. Do not disturb the low frequency components but check whether the high frequency components are expandable or not i.e., by embedding a bit(b) either 1 or 0 to form (Y (2n+1)) by
(Y (2n+1)) = 2 * (Y (2n+1)) + b
The odd and the even pixels formed after watermarking should lie in [0,255] for lossless recovery.
Restoring the original image
i. Calculate the (5, 3) wavelet transform of the received watermarked image.
ii. Collect all the LSB’s of the high frequency coefficients which have 1 in the location map to form the watermark bits.
iii. Use these high frequency coefficients and the original low frequency coefficients and apply (5,3) Inverse wavelet transform to get the reconstructed image.
iv. Check whether or not the reconstructed watermark is same as the original watermark.
4.2 Creating Graphical User Interfaces (GUIs)
GUIDE, the MATLAB Graphical User Interface development environment, provides a set of tools for creating graphical user interfaces (GUIs). These tools greatly simplify the process of designing and building GUIs.
You can use the GUIDE tools to
Lay out the GUI and to
Program the GUI
- Lay out the GUI
Using the GUIDE Layout Editor, you can lay out a GUI easily by clicking and dragging GUI components - such as panels, buttons, text fields, sliders, menus, and so on - into the layout area. Layout window created using Matlab GUIDE was shown in Figure 4.6
- Program the GUI
GUIDE automatically generates an M-file that controls how the GUI operates. The M-file initializes the GUI and contains a framework for all the GUI call backs - the commands that are executed when a user clicks a GUI component. Using the M-file editor, you can add code to the call backs to perform the functions you want them to.
M – file Editor
The functions get created by default after creating the layout, the most important factor to be considered here is to mention the appropriate ‘Tag’ and ‘String’ name of the considerable block or window.
1 Discrete Wavelet Transforms
The discrete wavelet transform (DWT) became a very versatile signal processing tool after Mallat [10] proposed the multi-resolution representation of signals based on wavelet decomposition. The method of multi-resolution is to represent a function (signal) with a collection of coefficients, each of which provides information about the position as well as the frequency of the signal (function). The advantage of the DWT over Fourier transformation is that it performs multi-resolution analysis of signals with localization both in time and frequency, popularly known as time-frequency localization. As a result, the DWT decomposes a digital signal into different sub bands so that the lower frequency sub bands have finer frequency resolution and coarser time resolution compared to the higher frequency sub bands. The DWT is being increasingly used for image compression due to the fact that the DWT supports features like progressive image transmission (by quality, by resolution), ease of compressed image manipulation] region of interest coding, etc. Because of these characteristics, the DWT is the basis of the new JPEG2000 image compression standard [11].
5.2 One dimensional DWT
Any signal is first applied to a pair of low-pass and high-pass filters. Then down sampling (i.e., neglecting the alternate coefficients) is applied to these filtered coefficients. The filter pair (h, g) which is used for decomposition is called analysis filter-bank and the filter pair which is used for reconstruction of the signal is called synthesis filter bank.(g`, h`).The output of the low pass filter after down sampling contains low frequency components of the signal which is approximate part of the original signal and the output of the high pass filter after down sampling contains the high frequency components which are called details (i.e., highly textured parts like edges) of the original signal.
This approximate part can still be further decomposed into low frequency and high frequency components.
3 Two-Dimensional DWT
One dimensional DWT can be easily extended to two dimensions which can be used for the transformation of two dimensional images. A two dimensional digital image which can be represented by a 2-D array X [m,n] with m rows and n columns, where m, n are positive integers. First, a one dimensional DWT is performed on rows to get low frequency L and high frequency H components of the image. Then, once again a one dimensional DWT is performed column wise on this intermediate result to form the final DWT coefficients LL, HL, LH, HH. These are called sub-bands.
The LL sub-band can be further decomposed into four sub-bands by following the above procedure. This process can continue to the required number of levels. This process is called multi level decomposition. A three level decomposition of the given digital image is as shown [12]. High pass and low pass filters are used to decompose the image first row-wise and then column wise. Similarly, the inverse DWT is applied which is just opposite to the forward DWT to get back the reconstructed image,
6.1 Conclusions
The main objective of the thesis is to find a reversible (lossless) watermarking algorithm (which can be used for sensitive applications like medical and military purposes) which can be integrated with the JPEG2000 standard. Instead of embedding watermark directly in pixels of the image where the distortion is large, watermark bits are embedded in the high frequency coefficients where the embedding capacity is more. Since (5, 3) Discrete wavelet transform is considered as the standard lossless transform for JPEG2000 compression standard, this (5, 3) Integer to Integer transform is used and the watermark is embedded reversibly using difference expansion. To increase the embedding capacity, more than two bits at a time are used to embed in the high frequency coefficients and reversibility is confirmed. But as the number of bits to be embedded increases the Peak Signal to Noise Ratio (PSNR) reduces.
The result for embedding up to 4 watermark bits per coefficient is presented (maximum embedding capacity over 90000 bits is achieved for a 256x256 lenna image). Instead of random bits as the watermark, a 128x128 bit signature is taken as watermark and embedded in a baboon image and exact recovery is also confirmed. All the above said results are confirmed by MATLAB simulations.
The complete VLSI implementation of watermark embedding is presented.
Hardware model
- Architecture for Reversible Watermarking have been designed in such a way that more than 1bit can be embedded in the pixel of an image by comparing each watermarked value with the four limits obtained for checking underflow and overflow conditions
- To retain image quality compression is avoided and differential expansion method is used for embedding the watermark
Software Reference model
- Tian’s algorithm of Difference Expansion has been considered as a reference algorithm for implementing Reversible watermarking algorithm
- (5, 3) Discrete wavelet transform is considered as the standard lossless transform for JPEG2000 compression standard, this (5, 3) Integer to Integer transform is used and the watermark is embedded reversibly using difference expansion
- By difference expanding the high pass coefficients of the image and embedding the watermark in those high pass coefficient, maximum embedding capacity over 90000 bits is achieved for a 256x256 lenna image
- A 128x128 bit signature is taken as watermark and embedded in a lenna image and exact recovery is confirmed after performing (5,3)Inverse Discrete Wavelet Transform
6.2 Future scope of the Work
As future work,
- Maximum payload size that can be embedded in the image can be determined before embedding the watermark.
- The hardware design for embedding multiple bits in a coefficient can be done.
- This watermarking block can be integrated with the other blocks of the JPEG2000 compression standard.
[1] Primo Braga, C. A., C. Fink, and C. Paz Sepulveda, Intellectual Property Rights and Economic Development, technical report, The World Bank, Washington D.C., 2000.
[2] I. Cox, M. L. Miller, and J. A. Bloom, Digital Watermarking. San Francisco , CA
[3] J. Tian, “Wavelet-based reversible watermarking for authentication”, in Proceedings of SPIE Sec. and Watermarking of Multimedia Cont. IV, vol. 4675, Jan 2002.
[4] Michael Arnold, Martin Schmucker and Stephen D. Wolthusen, “Techniques and Applications of Digital Watermarking and Content Protection”, published by Artech House, Boston , London
[5] Perry, B., B. MacIntosh, and D. Cushman, “Digimarc MediaBridge—The Birth of a Consumer Product, from Concept to Commercial Applicaton”, in E. J. Delp and P. W. Wong, (eds.), Proceedings of Electronic Imaging 2002, Security and Watermarking of Multimedia Contents IV, San Jose, CA, pp. 118–123, January 2002.
[6] Fridrich, J., and M. Goljan, “Practical Steganalysis of Digital Images: State of the Art”, in E. J. Delp and P. W. Wong, (eds.), Proceedings of Electronic Imaging, Security and Watermarking of Multimedia Contents IV, San Jose, CA, pp. 1–13, January 2002.
[7] Voloshynovskiy, S. V., et al., “StegoWall: Blind Statistical Detection of Hidden Data”, in P.W.Wong and E. J. Delp, (eds.), Proceedings of Electronic Imaging, Security andWatermarking of Multimedia Contents IV, San Jose, CA, pp. 57–68, January 2002.
[8] HyeRan Lee and KyungHyun Rhee, “Reversible Data Embedding for Tamper-Proof Watermarks”, Proceedings of the First International Conference on Innovative Computing, Information and Control, 2006
[9] Jun Tian, “Reversible Data Embedding Using a Difference Expansion”, IEEE Transactions on Circuits and Systems for Video Technology, vol.13, No.8, August 2003.
[10] ISO/IEC 15444-6, Final Committee Draft, “Information Technology-JPEG2000 Image Coding System, Part 6: Compound Image File Format,” 2001.
[11] ISO/IEC 15444-12, “Information Technology-JPEG2000 Image Coding System, Part-12: I S 0 Base Media File Format,” 2004.
[12] S. G. Mallat, “Multifrequency channel decomposition of images and wavelet models’’, IEEE Trans. Acoust., Speech, Signal Processing, vol. 37, pp. 2091-2110 December 1989.
[13] Jen-Bang Feng, Iuon-Chang Lin, Chwei-Shyong Tsai, and Yen-Ping Chu, “Reversible Watermarking: Current Status and Key Issues”, International Journal of Network Security, Vol.2, No.3, PP.161–171, May 2006 (http://isrc.nchu.edu.tw/ijns/)
[14] Wei Hsin Chang, Yew San Lee, Wen Shiaw Peng and Chen Yi Lee, “A Line-Based Memory Efficient and Programmable Architecture for 2D DWT using Lifting Scheme”, International Symposium on Circuits and Systems vol. 4 pp. 330–333, 2001.
[15] C. Chrysafis and A. Ortega,“Line-based, reduced memory, wavelet image compression”, IEEE Transaction on. Image Processing,, vol. 9, pp.378389, Mar. 2000.
[16] J.Tian, “High capacity reversible data embedding and content authentication,” in IEEE Proceedings of International Conference on Acoustics, Speech, and Signal Processing, vol. 3, pp. III–517–20, Apr 2003.
[17] A.Alattar, “Reversible watermark using the difference expansion of a generalized integer transform,” IEEE Transactions on Image Processing, vol. 13, no. 8, pp. 1147–1156, Aug
Posts
0 comments:
Post a Comment