Image Authentication
Content
TABLE OF CONTENTS
Page No. Acknowledgement Abstract 6 7
List Of Figures :Figure No 1(a) 1(b) 1(c) 4(a) 5(a) 5(b) 5(c) 5(d) 5(e) 5(f) Figure Name Image of Baboon Image of Lena Image of Peppers Elements of a Watermarking System Watermark-embedded image Watermark Visible Watermark Invisible Watermark Removing the embedded watermark from watermarked picture Detectable and Readable Watermarking Systems
Chapters : Chapter No. 1 2 3 4 5 6 7 Topic History Important Definitions Test Images used in Simulation Concepts of Digital Watermarking Types of Digital Watermarking Reversible Watermarking References
ABSTRACT
Watermarking has been studied extensively for many years, which is a technique for invisibly embedding information into a digital content, such as image, audio, or video data, as a way to trace ownership or prevent piracy. With the rapid development of digital technology, the treatments for digital data such as copyright protection and ownership demonstration are becoming more and more of greater importance. Due to this, the motivation of this project is to introduce some reversible watermarking methods for solving the problems encountered.
Most of the watermarking techniques are irreversible, that is to say, the image cannot return to its original state after embedding. In the embedding procedure, the irreversible distortion of the original content is introduced. Although this distortion is imperceptible, we will never regain the original content. This may not be acceptable to some sensitive applications such as military data, medical data and 2D-vector data for geographical information system(GIS). Reversible Watermarking is also called lossless watermarking i.e., the original content can be completely restored when decoding. Thus, it is an important requirement for many watermarking techniques.
CHAPTER - 1 : HISTORY of Water-marking
Watermarking is an old technique. It had been used widely in the past. A traditional and well-known example is the use of invisible ink. People wrote secret information using invisible ink in order to avoid detection from prying eyes. From a general point of view, the definition of watermarking may be thought of as a method to insert or embed extra information into the media, and also to indicate the method used to obtain the embedded information. In the past decade, owing to the rapid-development of computer technology, people had shifted their focus from traditional media to digital media. As a result, watermarking techniques for digital data have been developed and have become popular.
CHAPTER - 2 : Important Definitions
The general definitions of some common terms used in the area of watermarking are listed below -
1. Watermark : The information to be hidden. The term watermark also contains a hint that hidden information is transparent like water. 2. Cover Media : The media used for carrying the watermark. Sometimes the terms original media and the host media are used to express it. 3. Watermarked Data : The media which contains the watermark. 4. Embedding : The procedure used for inserting the watermark into the cover media. 5. Extraction : The procedure used for extracting the embedded watermark from the watermarked data. 6. Detection : The procedure used for detecting whether the given media containing a particular watermark. 7. Watermarking : The method which contains the embedding operator and the extraction/detection operator. 8. Noise : The natural noise occurred to the watermarked data during transmission.
9. Attack : The artificial processes used for modifying the watermarked data in order to destroy the watermark contained in the watermarked data. 10. Attacked Data : The watermarked data which contains natural noise and/or artificial modification.
CHAPTER - 3 : TEST IMAGES used in SIMULATION
The watermarking schemes discussed in this report are all image-based techniques. So, to test the performance of them, a number of experiments had been done. For convenience, the images used frequently are listed here.
The gray-scaled images shown in figure (1) were used as the cover images. We name them LENA, PEPPERS and BABOON respectively. The size of either image is 512 X 512 pixels.
Fig 1(a) The image of Baboon
Fig 1(b) The image of LENA
Fig 1(c) The image of PEPPERS
CHAPTER - 4 : Concepts of DIGITAL WATERMARKING
Elements of a Watermarking System A watermarking system is regarded as a communication system consisting of three main parts : a transmitter, a communication channel, and a receiver as illustrated below :-
Fig 4(a) Elements of a Watermarking System
The information-coding procedure encodes, compresses, and/or encrypts the original watermark W according to a user key K1. The data embedding procedure then embeds the encoded result Wc into the host data H according to another user key K2. The watermarked data HW is then delivered to the receiver via some kind of channel. During the transmission, some natural noise or artificial attacks may occur, as a result the received data H’W of the receiver may be different from the output data HW of the transmitter. To recover the information hidden in H’W, the hidden-data-extraction procedure is executed. In the above system, either K1, K2 or H may or may not have to be presented, according to the algorithms used.
CHAPTER - 5 : TYPES of Digital Watermarking
Types of Digital Watermarking Techniques (1) Visible and Invisible From the view point as to whether the embedded watermark can be seen by bare human eyes or not, all watermarking techniques can be classified as visible techniques and invisible techniques. For example, figure 5(c), shows a picture which contains a visible logo of University of South Australia in its top left corner and figure 5(d)
Fig 5(c)[Top], 5(d)[bottom]
An example of applying the visible watermarking technique and invisible watermarking technique to a given picture (a) The original picture (b) The watermark (c) The watermarked picture containing a visible logo in its top-left corner. (d) The watermarked picture containing an invisible watermark therein.
Disadvantages in visible watermarking techniques : (i) (ii) The visible watermark is not difficult to be removed. The visible watermark degrades the visual quality of the host picture.
In the invisible type of watermarking techniques, the embedded watermark is invisible. It is difficult to distinguish between the original image and the watermarked image. Thus, it is not easy to remove or destroy the embedded watermark without degrading the visual quality of the watermarked image significantly.
Fig 5(e) An example of removing the embedded watermark from the
watermarked picture
(2) Detectable and Readable
From the view point as to what kind of information can be obtained from the watermarked data, the watermarking techniques can be classified as detectable techniques and readable techniques. In the detectable types of techniques, one can only verify if a specific signal (the watermark) is contained in the cover work. In other words, the detectable type of systems only gives a binary answer : yes or no. In contrast, the readable watermarking systems extract and reveal the embedded watermark. Researchers use detection to illustrate the process of obtaining a binary answer, and extraction to express the process of revealing the hidden watermark. For those techniques which belong to the detectable type, the embedded watermark has to be presented during detection. This kind of technique is more private since it is impossible for an attacker to guess the content of the embedded watermark, especially if the embedded watermark is encrypted beforehand.
Fig 5(f) The detectable and the readable watermarking systems
(3) Spatial, Transform, and Quantisation
The question of where to hide the signal of the watermark may be divided into three categories : spatial-domain-based techniques, transform-domain-based techniques, and quantization-domain-based techniques. The main concept of spatial-domain-based techniques is to modify the raw data (pixels) of the original host image directly when hiding the watermark bits. The traditional method is to change the Last Significant Bits (LSB) of certain pixels of the host image according to the watermark bits. For transform-domain-based techniques, the raw data of the host image are first transformed into frequencies using the discrete cosine transform (DCT), the discrete wavelet transform (DWT), or other types of transforms. These frequencies are then modified according to the watermark bits so that the goal of data hiding can be achieved. Then, the inverse transform is executed and a watermarked image is formed. The quantization-domain-based techniques, such as vector quantization (VQ), first quantify the host image using the predefined code-vectors. The indices obtained are then modified according to the watermark bits. The
recovery process is finally performed to reconstruct a watermarked image from these modified lines.
Spatial-domain-based techniques : Advantages -easy implementation, better visual quality, and shorter coding time. Disadvantages -weak robustness
Transform-based-techniques : Advantages -better robustness and good visual quality in watermarked result. Disadvantages -They consume more time in the transform and inverse-transform procedures.
Quantisation-based-techniques : Advantages -They enhance the traditional quantization systems the watermarking ability
(4) Robust, Semi-Fragile, and Fragile
Watermarking techniques can also be classified as robust, semi-fragile, and fragile techniques, according to whether the techniques have strong resistance to natural noise and/or to artificial modification (named attack). If a watermarking technique can detect or extract the hidden watermark successfully from the watermarked data when noise and/or attack occurred, it is called a robust technique. In contrast, a
watermarking technique that cannot resist noise or attacks is called a fragile technique. There are some watermarking techniques which have strong resistance to some kinds of noise or attack but have weak resistance to other kinds of noise or attack. These techniques are named as semi-fragile techniques.
(5) Blind and Non-blind
If a watermarking technique resorts to the comparison between the original non-watermarked data and the watermarked one to recover the watermark, it can be classified as blind technique and non-blind technique. A blind watermarking technique requires no original data for detection or extraction. In contrast, a non-blind watermarking technique requires the original data to be presented during detection or extraction. In real world practices, non-blind watermarking algorithms are unsuitable for many practical applications in that they require the non-watermarked data to be presented during extraction or detection. Currently, most researchers are focusing on blind watermarking techniques rather than non-blind watermarking techniques. In addition, definitions of blind and non-blind nowadays have been extended. Some researchers think that if a watermarking technique requires the presence of any information used in embedding for watermark extraction, it then should be classified as non-blind. Based on this definition, one watermarking technique which requires no nonwatermarked data but requires the knowledge of embedding position when extraction, is regarded as a non-blind technique.
(6) Public and Private
A watermark is named private if only the authorized users can recover it. In other words, it is impossible for unauthorized people to extract the information hidden within the host data.
By contrast, a watermarking technique that allows anyone to read the embedded watermark is referred as a public watermarking technique. From the view point of information theory, security cannot be based on algorithms but rather on the choice of the user key. Therefore, it is believed that private watermarking techniques have superior robustness when compared to public watermarking techniques. (7) Symmetric and Asymmetric
A watermarking algorithm is called symmetric if the detection/extraction process makes use of the same set of parameters used in the embedding process. Here the parameters include the secret keys and other information which may be used to define the embedding position and the embedding process. And, a watermarking algorithm is said asymmetric if it uses different keys and parameters for the embedding and the detection/extraction operations. It is believed that for symmetric watermarking techniques a knowledge of these parameters is likely to give pirates enough information to remove the watermark from the watermarked data. Therefore, increasing attention to remove has been given to asymmetric watermarking schemes. Generally, asymmetric watermarking algorithms use a private key for watermark embedding and use a public key for watermark detection/extraction. (8) Reversible and Non-reversible
A watermarking algorithm is said reversible if the watermarked signal can be converted to a non-watermarked signal after the embedded watermark is extracted. Contrary, the watermarking algorithms that cannot convert the watermarked signal to a non-watermarked signal are named as nonreversible watermarking algorithms.
Currently, most of the existing watermarking algorithms are nonreversible algorithms, since the selected signals of the cover media have been changed permanently for carrying the watermark bits.
CHAPTER - 6 : REVERSIBLE WATERMARKING
In reversible watermarking, the watermark is embedded in a reversible way so that one can extract the hidden watermark and also restore the digital content to its original state.
History -The first publication to investigate reversible watermarking was by Barton in 1997. In his article, digital information is embedded within a digital stream for authentication. The embedded information can be extracted and used to verify the original digital data stream. Then Honsinger proposed a lossless data embedding technique that allows the recovery of the original image. The watermark is embedded using modulo arithmetic. Later on in Macq‟s article, a modification to Honsinger‟s algorithm is presented to achieve lossless watermarking. However, the embedded image with salt and pepper artifact appeared on both Honsinger and Macq‟s techniques. In order to reduce the artifact Vleeschouwer proposed a lossless watermarking algorithm by circular interpretation of bijective transforms. Besides, Fridrich developed a reversible data embedding algorithm based on compressing one of the least significant bit planes of the original image. They also described two extended reversible data embedded techniques for all image formats. Although these techniques can reduce the salt and pepper artifact, the capacity of the algorithms is low. Celik introduced a high capacity and low distortion reversible watermarking algorithm in 2002. The pixels of the original image are quantized and the residues are compressed using a lossless image compression algorithm in order to create capacity for the payload data.
Not long ago, Tian used a difference expansion method to reversibly embed a payload into digital images. He explored the redundancy in digital images to achieve a high capacity and low distortion reversible watermarking. His method divides the image into pairs of pixels, and some difference values that are not expected to cause an overflow or underflow are selected for the difference expansion (DE). One watermark bit will be embedded into the difference of each selected pixel pairs. Finally a location map of the selected expandable pixel pairs is losslessly compressed and included in the payload. In order to achieve a high capacity reversible data embedding method for digital images, a multiple embedding is employed in Tian‟s algorithm. In 2004, Alattar proposed a generalized difference expansion method. This method hides several bits in the difference expansion of vectors of adjacent pixels. Besides, Voigt proposed a reversible watermarking scheme for 2D-vector data. Only a few reversible schemes are designed for vector maps until now. A survey of digital vector map watermarking has been discussed .
Reversible Watermarking Algorithm - Flow Chart
Tian’s Algorithm -Most reversible watermarking algorithms rely on some forms of lossless compression to create space for embedding the payload. Tian used the redundancy in digital images to achieve very high embedding capacity and keeps the distortion low. He employed the difference expansion (DE) technique to reversibly embed watermarkinto digital images. Let‟s explain the DE with a simple example. Assume we have two neighbouring pixels with values p1=106 and p2=104. Then the diffence d and the average g can be computed as follows :d = p1 – p2 = 106 -104 = 2 g =[ p1 + p2 106 + 104 ]=[ ] = 105 2 2
Here, [ ] denotes the least integer. To embed a watermark bit w = 1 into the pixel pair, the difference d is represented using binary format, shift it left by one bit and append the watermark bit w into the vacant least significant bit (LSB). If l is the bit length of d (i.e., d = bl-1 bl-2 …..b0), then the new difference value d‟ can be obtained as :d‟= bl-1 bl-2 …..b0 w = 2 × d + w = 2 × 2 + 1 = 5 Finally, the new pixel values of p1‟ and p2‟ are computed as follows. d' + 1 p1‟ = g + [ 2 ] = 105 + 3 = 108 d' p2‟ = g - [ 2 ] = 105 – 2 = 103 In the decoder, the watermark bit can be extracted from the LSB of the difference value and the orognal difference value can be restored. d‟ = p1‟ – p2‟ = 108 – 103 = 5 w = LSB(d‟) = LSB(l0l2) = 1
d' 5 d = [ 2 ] = [2] = 2 Then the original pixel values p1 and p2 can be restored completely p1' + p2' d+1 p1 = [ 2 ] + [ 2 ] = 105 + 1 = 106 p1' + p2' d p2 = [ 2 ] - [2] = 105 - 1 = 104 To prevent the overflow and underflow problems in a gray scale image, the pixel values must satisfy d' + 1 0 ≤ g + [ 2 ] ≤ 255 d' 0 ≤ g - [ 2 ] ≤ 255 As the DE does not lead to overflow or underflow, the pixel pair is called expandable. From the above constraints, a generalized boundary condition is derived for the expandable pixel pairs. | 2 × d + w | ≤ min( 2 × (255 - g), 2 × g + 1) Besides, a pixel pair is called changeable if d | 2 × [2] + w | ≤ min( 2 × (255 - g), 2 × g + 1) It means that the new pixel values will not introduce an overflow or underflow after changing the LSB of the difference value. Obviously it can be seen that an expandable pixel pair is also changeable. And the conditions on expandable and changeable are equivalent when d = 0 or -1.
Now, the embedding procedure can be stated as follows :1. The original image is partitioned in pixel pairs firstly. The pairing could be horizontally or vertically. 2. Divide the set of pixel pairs into four subsets S1, S2 , S3 and S4. The subset S1 contains all expandable pixel pairs that are not in S1. The subset S3 contains all changeable pixel pairs that are not in S 1 U S2. The subset S4 contains all non changeable pixel pairs. To control the distortion between the original image and embedded image, a threshold T is set and the subset S2 is partitioned into S21 and S22. Here, S21 contains all pixel pairs with |d| ≤ T in S2 and S22 contains all pixel pairs with |d| > T in S2. 3. To identify the locations of the pixel pairs in S1, S21, S22, S3 and S4, a location map M is created. The symbol “l” in M indicates the locations of S1 or S21, and the symbol “0” indicates the locations of S22, S3 or S4.then the location map will be losslessly compressed by a JBIG2 compression or a run length coding. The compressed bitstream is denoted as B1. A unique identifier EOS is appended to B1 in the meantime. 4. Extract the LSBs of difference values in S22 and S3. Collect these bits into bitstream B2. 5. Assume the watermark (payload) to be embedded in B3. Concatenate B1, B2, B3 to form the bitstream B = B1 U B2 U B3 = b1b2….bm , where m is the bitlength of B. finally the bitstream B is embedded in the host image.
To extract the watermark and restore the original image, the decoding procedure is summarized in the following steps :-
1. Extract the LSBs of the difference values of all changeable pixel pairs. Collect these bits into bitstream B. 2. Retrieve B1 from B and decompress it to restore the location map M.
3. Determine the expandable pixel pairs using M and restore the original pixel pairs. 4. Retrieve B2 from B and the other changeable pixel pairs can be restored.
5. Extract the embedded watermark from B.
CHAPTER - 7 : REFERENCES
1) Innovations in Digital Watermarking Techniques by Feng-Hsing Wang, Jeng-Shyang Pan, and Lakhmi C.Jain (Publisher-Springer) 2) Paper on „An extended difference expansion algorithm for reversible watermarking‟ by Hsien-Wen Tseng, and Chin-Chen Chang 3) Paper on „Reversible Watermarking by Difference Expansion‟ by Jun Tian 4) Paper on „Reversible Watermarking using Modified Difference Expansion‟ by Mohd. K. Yaqub, and Ahmed Al-Jaber 5) Review Article on „Reversible Watermarking Techniques : An overview and a Classification‟ by Roberto Caldelli, Francesco Filippini, and Rudy Becarelli 6) Paper on „Securing Biometric Images using Reversible Watermarking‟ by Sabu M.Thampi, and Ann Jisma Jacob
7)Sources on Internet - many websites like „prepressure.com‟ ,
„fileformat.info‟ etc.
Page No. Acknowledgement Abstract 6 7
List Of Figures :Figure No 1(a) 1(b) 1(c) 4(a) 5(a) 5(b) 5(c) 5(d) 5(e) 5(f) Figure Name Image of Baboon Image of Lena Image of Peppers Elements of a Watermarking System Watermark-embedded image Watermark Visible Watermark Invisible Watermark Removing the embedded watermark from watermarked picture Detectable and Readable Watermarking Systems
Chapters : Chapter No. 1 2 3 4 5 6 7 Topic History Important Definitions Test Images used in Simulation Concepts of Digital Watermarking Types of Digital Watermarking Reversible Watermarking References
ABSTRACT
Watermarking has been studied extensively for many years, which is a technique for invisibly embedding information into a digital content, such as image, audio, or video data, as a way to trace ownership or prevent piracy. With the rapid development of digital technology, the treatments for digital data such as copyright protection and ownership demonstration are becoming more and more of greater importance. Due to this, the motivation of this project is to introduce some reversible watermarking methods for solving the problems encountered.
Most of the watermarking techniques are irreversible, that is to say, the image cannot return to its original state after embedding. In the embedding procedure, the irreversible distortion of the original content is introduced. Although this distortion is imperceptible, we will never regain the original content. This may not be acceptable to some sensitive applications such as military data, medical data and 2D-vector data for geographical information system(GIS). Reversible Watermarking is also called lossless watermarking i.e., the original content can be completely restored when decoding. Thus, it is an important requirement for many watermarking techniques.
CHAPTER - 1 : HISTORY of Water-marking
Watermarking is an old technique. It had been used widely in the past. A traditional and well-known example is the use of invisible ink. People wrote secret information using invisible ink in order to avoid detection from prying eyes. From a general point of view, the definition of watermarking may be thought of as a method to insert or embed extra information into the media, and also to indicate the method used to obtain the embedded information. In the past decade, owing to the rapid-development of computer technology, people had shifted their focus from traditional media to digital media. As a result, watermarking techniques for digital data have been developed and have become popular.
CHAPTER - 2 : Important Definitions
The general definitions of some common terms used in the area of watermarking are listed below -
1. Watermark : The information to be hidden. The term watermark also contains a hint that hidden information is transparent like water. 2. Cover Media : The media used for carrying the watermark. Sometimes the terms original media and the host media are used to express it. 3. Watermarked Data : The media which contains the watermark. 4. Embedding : The procedure used for inserting the watermark into the cover media. 5. Extraction : The procedure used for extracting the embedded watermark from the watermarked data. 6. Detection : The procedure used for detecting whether the given media containing a particular watermark. 7. Watermarking : The method which contains the embedding operator and the extraction/detection operator. 8. Noise : The natural noise occurred to the watermarked data during transmission.
9. Attack : The artificial processes used for modifying the watermarked data in order to destroy the watermark contained in the watermarked data. 10. Attacked Data : The watermarked data which contains natural noise and/or artificial modification.
CHAPTER - 3 : TEST IMAGES used in SIMULATION
The watermarking schemes discussed in this report are all image-based techniques. So, to test the performance of them, a number of experiments had been done. For convenience, the images used frequently are listed here.
The gray-scaled images shown in figure (1) were used as the cover images. We name them LENA, PEPPERS and BABOON respectively. The size of either image is 512 X 512 pixels.
Fig 1(a) The image of Baboon
Fig 1(b) The image of LENA
Fig 1(c) The image of PEPPERS
CHAPTER - 4 : Concepts of DIGITAL WATERMARKING
Elements of a Watermarking System A watermarking system is regarded as a communication system consisting of three main parts : a transmitter, a communication channel, and a receiver as illustrated below :-
Fig 4(a) Elements of a Watermarking System
The information-coding procedure encodes, compresses, and/or encrypts the original watermark W according to a user key K1. The data embedding procedure then embeds the encoded result Wc into the host data H according to another user key K2. The watermarked data HW is then delivered to the receiver via some kind of channel. During the transmission, some natural noise or artificial attacks may occur, as a result the received data H’W of the receiver may be different from the output data HW of the transmitter. To recover the information hidden in H’W, the hidden-data-extraction procedure is executed. In the above system, either K1, K2 or H may or may not have to be presented, according to the algorithms used.
CHAPTER - 5 : TYPES of Digital Watermarking
Types of Digital Watermarking Techniques (1) Visible and Invisible From the view point as to whether the embedded watermark can be seen by bare human eyes or not, all watermarking techniques can be classified as visible techniques and invisible techniques. For example, figure 5(c), shows a picture which contains a visible logo of University of South Australia in its top left corner and figure 5(d)
Fig 5(c)[Top], 5(d)[bottom]
An example of applying the visible watermarking technique and invisible watermarking technique to a given picture (a) The original picture (b) The watermark (c) The watermarked picture containing a visible logo in its top-left corner. (d) The watermarked picture containing an invisible watermark therein.
Disadvantages in visible watermarking techniques : (i) (ii) The visible watermark is not difficult to be removed. The visible watermark degrades the visual quality of the host picture.
In the invisible type of watermarking techniques, the embedded watermark is invisible. It is difficult to distinguish between the original image and the watermarked image. Thus, it is not easy to remove or destroy the embedded watermark without degrading the visual quality of the watermarked image significantly.
Fig 5(e) An example of removing the embedded watermark from the
watermarked picture
(2) Detectable and Readable
From the view point as to what kind of information can be obtained from the watermarked data, the watermarking techniques can be classified as detectable techniques and readable techniques. In the detectable types of techniques, one can only verify if a specific signal (the watermark) is contained in the cover work. In other words, the detectable type of systems only gives a binary answer : yes or no. In contrast, the readable watermarking systems extract and reveal the embedded watermark. Researchers use detection to illustrate the process of obtaining a binary answer, and extraction to express the process of revealing the hidden watermark. For those techniques which belong to the detectable type, the embedded watermark has to be presented during detection. This kind of technique is more private since it is impossible for an attacker to guess the content of the embedded watermark, especially if the embedded watermark is encrypted beforehand.
Fig 5(f) The detectable and the readable watermarking systems
(3) Spatial, Transform, and Quantisation
The question of where to hide the signal of the watermark may be divided into three categories : spatial-domain-based techniques, transform-domain-based techniques, and quantization-domain-based techniques. The main concept of spatial-domain-based techniques is to modify the raw data (pixels) of the original host image directly when hiding the watermark bits. The traditional method is to change the Last Significant Bits (LSB) of certain pixels of the host image according to the watermark bits. For transform-domain-based techniques, the raw data of the host image are first transformed into frequencies using the discrete cosine transform (DCT), the discrete wavelet transform (DWT), or other types of transforms. These frequencies are then modified according to the watermark bits so that the goal of data hiding can be achieved. Then, the inverse transform is executed and a watermarked image is formed. The quantization-domain-based techniques, such as vector quantization (VQ), first quantify the host image using the predefined code-vectors. The indices obtained are then modified according to the watermark bits. The
recovery process is finally performed to reconstruct a watermarked image from these modified lines.
Spatial-domain-based techniques : Advantages -easy implementation, better visual quality, and shorter coding time. Disadvantages -weak robustness
Transform-based-techniques : Advantages -better robustness and good visual quality in watermarked result. Disadvantages -They consume more time in the transform and inverse-transform procedures.
Quantisation-based-techniques : Advantages -They enhance the traditional quantization systems the watermarking ability
(4) Robust, Semi-Fragile, and Fragile
Watermarking techniques can also be classified as robust, semi-fragile, and fragile techniques, according to whether the techniques have strong resistance to natural noise and/or to artificial modification (named attack). If a watermarking technique can detect or extract the hidden watermark successfully from the watermarked data when noise and/or attack occurred, it is called a robust technique. In contrast, a
watermarking technique that cannot resist noise or attacks is called a fragile technique. There are some watermarking techniques which have strong resistance to some kinds of noise or attack but have weak resistance to other kinds of noise or attack. These techniques are named as semi-fragile techniques.
(5) Blind and Non-blind
If a watermarking technique resorts to the comparison between the original non-watermarked data and the watermarked one to recover the watermark, it can be classified as blind technique and non-blind technique. A blind watermarking technique requires no original data for detection or extraction. In contrast, a non-blind watermarking technique requires the original data to be presented during detection or extraction. In real world practices, non-blind watermarking algorithms are unsuitable for many practical applications in that they require the non-watermarked data to be presented during extraction or detection. Currently, most researchers are focusing on blind watermarking techniques rather than non-blind watermarking techniques. In addition, definitions of blind and non-blind nowadays have been extended. Some researchers think that if a watermarking technique requires the presence of any information used in embedding for watermark extraction, it then should be classified as non-blind. Based on this definition, one watermarking technique which requires no nonwatermarked data but requires the knowledge of embedding position when extraction, is regarded as a non-blind technique.
(6) Public and Private
A watermark is named private if only the authorized users can recover it. In other words, it is impossible for unauthorized people to extract the information hidden within the host data.
By contrast, a watermarking technique that allows anyone to read the embedded watermark is referred as a public watermarking technique. From the view point of information theory, security cannot be based on algorithms but rather on the choice of the user key. Therefore, it is believed that private watermarking techniques have superior robustness when compared to public watermarking techniques. (7) Symmetric and Asymmetric
A watermarking algorithm is called symmetric if the detection/extraction process makes use of the same set of parameters used in the embedding process. Here the parameters include the secret keys and other information which may be used to define the embedding position and the embedding process. And, a watermarking algorithm is said asymmetric if it uses different keys and parameters for the embedding and the detection/extraction operations. It is believed that for symmetric watermarking techniques a knowledge of these parameters is likely to give pirates enough information to remove the watermark from the watermarked data. Therefore, increasing attention to remove has been given to asymmetric watermarking schemes. Generally, asymmetric watermarking algorithms use a private key for watermark embedding and use a public key for watermark detection/extraction. (8) Reversible and Non-reversible
A watermarking algorithm is said reversible if the watermarked signal can be converted to a non-watermarked signal after the embedded watermark is extracted. Contrary, the watermarking algorithms that cannot convert the watermarked signal to a non-watermarked signal are named as nonreversible watermarking algorithms.
Currently, most of the existing watermarking algorithms are nonreversible algorithms, since the selected signals of the cover media have been changed permanently for carrying the watermark bits.
CHAPTER - 6 : REVERSIBLE WATERMARKING
In reversible watermarking, the watermark is embedded in a reversible way so that one can extract the hidden watermark and also restore the digital content to its original state.
History -The first publication to investigate reversible watermarking was by Barton in 1997. In his article, digital information is embedded within a digital stream for authentication. The embedded information can be extracted and used to verify the original digital data stream. Then Honsinger proposed a lossless data embedding technique that allows the recovery of the original image. The watermark is embedded using modulo arithmetic. Later on in Macq‟s article, a modification to Honsinger‟s algorithm is presented to achieve lossless watermarking. However, the embedded image with salt and pepper artifact appeared on both Honsinger and Macq‟s techniques. In order to reduce the artifact Vleeschouwer proposed a lossless watermarking algorithm by circular interpretation of bijective transforms. Besides, Fridrich developed a reversible data embedding algorithm based on compressing one of the least significant bit planes of the original image. They also described two extended reversible data embedded techniques for all image formats. Although these techniques can reduce the salt and pepper artifact, the capacity of the algorithms is low. Celik introduced a high capacity and low distortion reversible watermarking algorithm in 2002. The pixels of the original image are quantized and the residues are compressed using a lossless image compression algorithm in order to create capacity for the payload data.
Not long ago, Tian used a difference expansion method to reversibly embed a payload into digital images. He explored the redundancy in digital images to achieve a high capacity and low distortion reversible watermarking. His method divides the image into pairs of pixels, and some difference values that are not expected to cause an overflow or underflow are selected for the difference expansion (DE). One watermark bit will be embedded into the difference of each selected pixel pairs. Finally a location map of the selected expandable pixel pairs is losslessly compressed and included in the payload. In order to achieve a high capacity reversible data embedding method for digital images, a multiple embedding is employed in Tian‟s algorithm. In 2004, Alattar proposed a generalized difference expansion method. This method hides several bits in the difference expansion of vectors of adjacent pixels. Besides, Voigt proposed a reversible watermarking scheme for 2D-vector data. Only a few reversible schemes are designed for vector maps until now. A survey of digital vector map watermarking has been discussed .
Reversible Watermarking Algorithm - Flow Chart
Tian’s Algorithm -Most reversible watermarking algorithms rely on some forms of lossless compression to create space for embedding the payload. Tian used the redundancy in digital images to achieve very high embedding capacity and keeps the distortion low. He employed the difference expansion (DE) technique to reversibly embed watermarkinto digital images. Let‟s explain the DE with a simple example. Assume we have two neighbouring pixels with values p1=106 and p2=104. Then the diffence d and the average g can be computed as follows :d = p1 – p2 = 106 -104 = 2 g =[ p1 + p2 106 + 104 ]=[ ] = 105 2 2
Here, [ ] denotes the least integer. To embed a watermark bit w = 1 into the pixel pair, the difference d is represented using binary format, shift it left by one bit and append the watermark bit w into the vacant least significant bit (LSB). If l is the bit length of d (i.e., d = bl-1 bl-2 …..b0), then the new difference value d‟ can be obtained as :d‟= bl-1 bl-2 …..b0 w = 2 × d + w = 2 × 2 + 1 = 5 Finally, the new pixel values of p1‟ and p2‟ are computed as follows. d' + 1 p1‟ = g + [ 2 ] = 105 + 3 = 108 d' p2‟ = g - [ 2 ] = 105 – 2 = 103 In the decoder, the watermark bit can be extracted from the LSB of the difference value and the orognal difference value can be restored. d‟ = p1‟ – p2‟ = 108 – 103 = 5 w = LSB(d‟) = LSB(l0l2) = 1
d' 5 d = [ 2 ] = [2] = 2 Then the original pixel values p1 and p2 can be restored completely p1' + p2' d+1 p1 = [ 2 ] + [ 2 ] = 105 + 1 = 106 p1' + p2' d p2 = [ 2 ] - [2] = 105 - 1 = 104 To prevent the overflow and underflow problems in a gray scale image, the pixel values must satisfy d' + 1 0 ≤ g + [ 2 ] ≤ 255 d' 0 ≤ g - [ 2 ] ≤ 255 As the DE does not lead to overflow or underflow, the pixel pair is called expandable. From the above constraints, a generalized boundary condition is derived for the expandable pixel pairs. | 2 × d + w | ≤ min( 2 × (255 - g), 2 × g + 1) Besides, a pixel pair is called changeable if d | 2 × [2] + w | ≤ min( 2 × (255 - g), 2 × g + 1) It means that the new pixel values will not introduce an overflow or underflow after changing the LSB of the difference value. Obviously it can be seen that an expandable pixel pair is also changeable. And the conditions on expandable and changeable are equivalent when d = 0 or -1.
Now, the embedding procedure can be stated as follows :1. The original image is partitioned in pixel pairs firstly. The pairing could be horizontally or vertically. 2. Divide the set of pixel pairs into four subsets S1, S2 , S3 and S4. The subset S1 contains all expandable pixel pairs that are not in S1. The subset S3 contains all changeable pixel pairs that are not in S 1 U S2. The subset S4 contains all non changeable pixel pairs. To control the distortion between the original image and embedded image, a threshold T is set and the subset S2 is partitioned into S21 and S22. Here, S21 contains all pixel pairs with |d| ≤ T in S2 and S22 contains all pixel pairs with |d| > T in S2. 3. To identify the locations of the pixel pairs in S1, S21, S22, S3 and S4, a location map M is created. The symbol “l” in M indicates the locations of S1 or S21, and the symbol “0” indicates the locations of S22, S3 or S4.then the location map will be losslessly compressed by a JBIG2 compression or a run length coding. The compressed bitstream is denoted as B1. A unique identifier EOS is appended to B1 in the meantime. 4. Extract the LSBs of difference values in S22 and S3. Collect these bits into bitstream B2. 5. Assume the watermark (payload) to be embedded in B3. Concatenate B1, B2, B3 to form the bitstream B = B1 U B2 U B3 = b1b2….bm , where m is the bitlength of B. finally the bitstream B is embedded in the host image.
To extract the watermark and restore the original image, the decoding procedure is summarized in the following steps :-
1. Extract the LSBs of the difference values of all changeable pixel pairs. Collect these bits into bitstream B. 2. Retrieve B1 from B and decompress it to restore the location map M.
3. Determine the expandable pixel pairs using M and restore the original pixel pairs. 4. Retrieve B2 from B and the other changeable pixel pairs can be restored.
5. Extract the embedded watermark from B.
CHAPTER - 7 : REFERENCES
1) Innovations in Digital Watermarking Techniques by Feng-Hsing Wang, Jeng-Shyang Pan, and Lakhmi C.Jain (Publisher-Springer) 2) Paper on „An extended difference expansion algorithm for reversible watermarking‟ by Hsien-Wen Tseng, and Chin-Chen Chang 3) Paper on „Reversible Watermarking by Difference Expansion‟ by Jun Tian 4) Paper on „Reversible Watermarking using Modified Difference Expansion‟ by Mohd. K. Yaqub, and Ahmed Al-Jaber 5) Review Article on „Reversible Watermarking Techniques : An overview and a Classification‟ by Roberto Caldelli, Francesco Filippini, and Rudy Becarelli 6) Paper on „Securing Biometric Images using Reversible Watermarking‟ by Sabu M.Thampi, and Ann Jisma Jacob
7)Sources on Internet - many websites like „prepressure.com‟ ,
„fileformat.info‟ etc.
