Digital Watermarking

A number of analogies to cryptographic concepts have been made about watermarking. In this paper, we argue that these analogies are misleading or incorrect, and highlight several analogies to support our argument. We believe that the fundamental role of watermarking is the reliable embedding and detection of information and should therefore be considered a form of communications. We note that the fields of communications and cryptography are quite distinct and while communications

systems

often combine technologies from the two fields, a layered architecture is applied that requires no knowledge of the layers above. We discuss how this layered approach can be applied to watermarking applications.

In this paper we introduce the problem of watermark security for systems in which an implementation of a watermark detector is available to an attacker. This paper serves as an introduction to a keynote talk at IWDW 2006. This talk will review two homomorphic encryption methods, viz. Paillier and Goldwasser & Micali, and their application to secure detection of Quantization Index Modulation (QIM) watermarks.

Steganography is often combined with cryptographic mechanisms. This enhances steganography by valuable properties that are originally left to cryptographic systems.

However, new problems for cryptographic mechanisms arise from the context of steganography. There are two sorts of steganographic tools: commercial tools with insecure or badly implemented cryptography and academic proof-of-concepts that abstain from the actual implementation of the cryptographic part.

Comparably to cryptography, steganography evolves in an iterative process of designing and breaking new methods. In this paper we examine the encoding properties and cryptographic functionality of steganographic tools to enable the detection of embedded information in steganograms even if the embedding part was otherwise secure.

In this paper, we propose a new method for the detection of LSB embedding in JPEG images. We are motivated by a need to further research the idea of the chi-square attack. The new method simply use the first-order statistics of DCT coefficients, but is more powerful to detect the random embedding in JPEG images. For evaluation, we used versions of Jsteg and Jphide with randomized embedding path to generate stego images in our experiments. In results, the proposed method outperforms the method of Zhang and Ping and is applicable to Jphide. The detection power of both proposed methods is compared to the blind classifier by Fridrich that uses 23 DCT features.

This paper presents a novel steganalysis scheme with high-dimensional feature vectors derived from co-occurrence matrix in either spatial domain or JPEG coefficient domain, which is sensitive to data embedding process. The class-wise non-principal components analysis (CNPCA) is proposed to solve the problem of the classification in the high-dimensional feature vector space. The experimental results have demonstrated that the proposed scheme outperforms the existing steganalysis techniques in attacking the commonly used steganographic schemes applied to spatial domain (Spread-Spectrum, LSB, QIM) or JPEG domain (OutGuess, F5, Model-Based).

This paper addresses a novel steganography method for images. Most statistical steganalysis algorithms are strong to defeat previous steganography algorithms. RS steganalysis and pixel difference histogram analysis are two well-known statistical steganalysis algorithms which detect non-random changes caused by embedding a secret message into cover image. In this paper, we first explain how two steganalysis algorithms exploit the effect of the non-random changes and then propose a new steganography method that avoids the non-random changes to evade statistical analysis methods. For this purpose, we adjust the embedding process to be more adaptive to cover image by considering embedding in Gray code bit planes, not natural binary bit planes, of cover images, and two parameters: (1) similarity threshold for selecting non-flat area in lower bit planes, and (2) size of flat blocks

n

×

n

in embedding bit planes. Experimental results show that the secret messages embedded by our method are undetectable under RS steganalysis and pixel difference histogram analysis.

This paper proposes a reversible watermarking scheme for error diffused halftone images. It exploits statistical features of 2×2 binary patterns in halftone images to embed data. According to a small look-up table constructed in advance, a state sequence is extracted and losslessly compressed, and the saved space is filled up with the watermark and some side information. We modulate the extracted state sequence into a new concatenated sequence by similar pair toggling, and meanwhile the watermark and the LUT are embedded. The proposed scheme can provide a considerable capacity and the original image can be recovered if its watermarked version is intact.

In this paper we present a print-scan resilient method to embed multibit messages into images. Multilevel watermarking principles are applied in order to embed a reference watermark and the message bits. Methods to embed a robust spatial domain reference watermark, utilizing HVS are proposed and methods to improve the estimation of affine transformation parameters from a periodic reference watermark are considered. The multibit message is embedded on the approximation coefficients of the wavelet transform, utilizing JND profile estimation and additive spread spectrum techniques. A blind correlation based detection method is made use of to recover the message. In the experiments the imperceptuality issues related to multilevel watermarking are considered with different parameter settings and the robustness against print-scan attack and JPEG compression is measured and the results are shown.

In this paper we propose a multi-set modulation technique to increase the hiding capacity within a binary document image. As part of this technique we propose an Automatic Threshold Calculation and Threshold Buffering, Shifted Space Distribution and Letter Space Compensation technique. The Automatic Threshold Calculation is used to distinguish word spaces from letter spaces. The Threshold Buffering is used to reduce the chance of misinterpretation of spaces during the detection phase, following printing and scanning. The Shifted Space Distribution and Letter Space Compensation techniques robustly embed a watermark into the binary document image. The Automatic Threshold Calculation has been shown to be successful in identifying word spaces for different types of fonts and font sizes. The combination of the Shifted Space Distribution, Letter Space Compensation and Threshold Buffering techniques have been shown to create a watermark that is robust to printing and scanning.

This paper presents a new multibit watermarking method. The method uses multiple orthonormalized watermark patterns of the same size as the host signal. In particular, the elements of each watermark pattern follow independent normal distribution. Each bit of the transmitted message is hidden using the dither quantizers to modify the projection of the host signal onto its corresponding watermark pattern. As a result, every hidden bit is spread over all elements of the host data and the extracting procedure is blind. Meanwhile, we consider how to choose a suitable quantization step size under the given distortion constraint. It is also proved mathematically that the upper bound of the bit error probability of our method is equal to one of the spread-transform dithered modulation (STDM) under the same situations. Experimental results show our scheme performs better than STDM in terms of bit error probability.

As most previous

wavelet analysis

based

3-D mesh watermarking

methods embed the watermark information into wavelet coefficients arranged in a certain order, they have not been used as

blind schemes

since the connectivity information must be exactly known in the watermark extraction process. In this paper, we propose a blind watermarking method based on wavelet analysis for 3-D mesh model. Two new techniques are introduced. One is to exploit the

statistical features

of

scale coefficients

on an approximation (low resolution) level for watermark embedding. Another is to extract the hidden watermark, not from the same resolution level as used in embedding process, but directly from the spatial domain. As the proposed watermark detection does not require the wavelet analysis, any pre-processing such as registration and re-sampling, is not needed. These techniques allow to detect the watermark without referring to the original meshes. In addition, the proposed are applicable directly to

irregular meshes

by using irregular wavelet analysis. Through simulations, we prove that our method is fairly robust against various attacks including topological ones.

This paper presents a novel watermarking scheme for 3D keyframe animation based on geometric structure and interpolator. The geometric watermarking embeds the watermark into the distribution of vertex coordinates in each of initial mesh models. The interpolator watermarking embeds the same watermark into key values in position interpolator. Experimental results show that the proposed scheme has the robustness against various geometric attacks and timeline attacks as well as the invisibility.

A robust video watermarking scheme via temporal segmentation and middle frequency component adaptive modification is presented. First, the video clip is sliced into a shot sequence by temporal video segmentation. Then, several consecutive shots are selected from the shot sequence to makeup a video segment with proper duration. The watermark is embedded into the video segment by equally dividing it into many units, and modifying the middle frequency component of frames in each unit. In addition, the scheme adjusts the modification strength of every coefficient adaptively according to the change between consecutive frames and Watson’s DCT based visual model. Experimental results show that the scheme is resilient to many kinds of temporal desynchronization, spatial desynchronization and photometric distortion at the same time.

One of the main goals of watermarking is to optimize capacity while preserving high video fidelity. This paper describes a watermarking scheme based on the spread spectrum paradigm with capacity enhancement using a state-of-the-art error correction technique – duo-binary turbo coding. A new watermark composition with novel bit-wise watermark bits interleaving scheme and bit-rate control on the macro-block level is proposed. In previous works, the perceptual watermark adjustment was mainly based on Watson Just Noticeable Difference (JND) model. A new JND estimation model based on block classification is presented. In addition, experimental results on perceptibility and robustness to transcoding are reported.

Image splicing is a commonly used technique in image tampering. This paper presents a novel approach to passive detection of image splicing. In the proposed scheme, the image splicing detection problem is tackled as a twoclass classification problem under the pattern recognition framework. Considering the high non-linearity and non-stationarity nature of image splicing operation, a recently developed Hilbert-Huang transform (HHT) is utilized to generate features for classification. Furthermore, a well established statistical natural image model based on moments of characteristic functions with wavelet decomposition is employed to distinguish the spliced images from the authentic images. We use support vector machine (SVM) as the classifier. The initial experimental results demonstrate that the proposed scheme outperforms the prior arts.

In this paper, we propose a new encryption method for intellectual property rights management of MPEG-4 coder. DCT coefficients and motion vectors (MVs) are used for the combination encryption of the MPEG-4 video. Experimental results indicate that the proposed joint and partial encryption technique provides levels of security and achieves a simple and coding-efficient architecture with no adverse impact on error resilience.

This paper presents a data hiding technique based on a new compression enhancement called Film Grain Technology. Film grain is a mid-frequency noise-like pattern naturally appearing in imagery captured on film. The Film Grain Technology is a method for modeling and removing the film grain, thus enhancing the compression efficiency, and then using the model parameters to create synthetic film grain at the decoder. We propose slight modifications to the decoder that enable the synthetic film grain to represent metadata available at the decoder. We examine a number of implementation approaches and report results of fidelity and robustness experiments.

Since digital right management of digital media data has received considerable attention recently, protection of halftone image documents becomes another important topic. Image-based visual cryptography is found to provide an alternative for applications of copyright protection by overlapping more than one secret embedded image to show the hidden information. In this paper, we propose a novel screening halftoning-based visual cryptography method for halftone image protection. Compared with the existing methods, the major contributions of our method contain (i) improved quality of the halftone images and extracted secrets; (ii) unlimited database size of protected halftone images; (iii) more than two halftone images can be overlapped to show the hidden secret; (iv) only one conjugate screen pair in our method is able to achieve the maximum clarity of extracted secrets in random screening. Experimental results and comparisons with a state of the art method demonstrate the effectiveness of our method.

Extensive testing shows that the audio Zernike moments in lower orders are very robust to common signal processing operations, such as MP3 compression, low-pass filtering, etc. Based on the observations, in this paper, a robust watermark scheme is proposed by embedding the bits into the low-order moments. By analyzing and deducting the linear relationship between the audio amplitude and moments, watermarking the low-order moments is achieved in time domain by scaling sample values directly. Thus, the degradation in audio reconstruction from a limited number of watermarked Zernike moments is avoided. Experimental works show that the proposed algorithm achieves strong robustness performance due to the superiorities of exploited low-order moments.

The quantization-based watermarking schemes such as QIM or SCS are known to be very vulnerable to the amplitude modification attack. The amplitude modification attack results in the change of quantization step size so the estimation of a modified quantization step size is required before watermark detection. In this paper, we analyze the quantization error function of the audio signal having any shape of probability density function, and analytically determine the search interval that minimizes the quantization error considering both detection performance and computational complexity. It is shown that the appropriate search interval can be determined from the frame-based mean and variance of the input signal without regard to its shape of probability density function. Experimental results for real audio data verify that the derived search interval provides the accurate estimation of the modified quantization step size under amplitude modification attack.

We present in this paper a new approach for universal JPEG steganalysis and propose studying statistics of the compressed DCT coefficients. This approach is motivated by the

Avalanche Criterion

of the JPEG lossless compression step. This criterion makes possible the design of detectors whose detection rates are independent of the payload. We design an universal steganalytic scheme using blocks of the JPEG file binary output stream. We compute higher order statistics over their Hamming weights and combined them with a Kullbak-Leibler distance between the probability density function of these weights and a benchmark one. We evaluate the universality of our detector through its capacity to efficiently detect the use of a new algorithm not used during the training step. To that goal, we examinate training sets produced by Outguess, F5 and JPhide-and-Seek. The experimental results we obtained show that our scheme is able to detect the use of new algorithms with high detection rate (≈90%) even with very low embedding rates (<10

− − 5

).

In this paper, using a new steganalytic method, namely the left-and-right cube analysis (LRCA), we explore the steganography exploiting the JPEG2000 image as the carrier medium, where the steganography works by substituting message bits for the least significant bits (LSBs) of the quantized wavelet coefficients. For the achievement of the LRCA implementation on JPEG2000 images, we establish an appropriate sampling rule to extract the sample vectors from the image data,

i.e.,

the quantized wavelet coefficients, in the wavelet domain. In experiments, we used the LSB steganography and the bit-plane complexity segmentation (BPCS) steganography in order to generate the stego images. In results, the proposed method works well for both steganographic methods, and outperforms some other previous steganalytic methods.

In 2005, Ginger Myles and Hongxia Jin proposed a software watermarking scheme based on converting

jump instructions

or

unconditional branch statements

(UBSs) by calls to a

fingerprint branch function

(FBF) that computes the correct target address of the UBS as a function of the generated fingerprint and integrity check. If the program is tampered with, the fingerprint and integrity checks change and the target address will not be computed correctly. In this paper, we present an attack based on tracking stack pointer modifications to break the scheme and provide implementation details. The key element of the attack is to remove the fingerprint and integrity check generating code from the program after disassociating the target address from the fingerprint and integrity value. Using the debugging tools that give vast control to the attacker to track stack pointer operations, we perform both subtractive and watermark replacement attacks. The major steps in the attack are automated resulting in a fast and low-cost attack.

To enable copyright protection and authentication, robust digital watermark can be embedded into multimedia contents imperceptibly. However, geometric distortions pose a significant threat to robust image watermarking because it can desynchronize the watermark information while preserving the visual quality. To overcome this, we developed an invariant domain with three transforms; Fast Fourier Transform (FFT), Log-Polar Mapping (LPM), and Dual Tree-Complex Wavelet Transform (DT-CWT). Shift invariance is obtained using FFT. Rotation and scaling invariance are achieved by taking the DT-CWT of a LPM output. Unlike most invariant schemes, our method eliminates explicit re-synchronization. The method resists geometric distortions at both global and local scales. It is also robust against JPEG compression and common image processing. In addition, it exploits perceptual masking property of the DT-CWT subbands, and its watermark detection step does not require the cover image. Experiment on a large set of natural images shows the robustness of the new scheme.

Till now, few desynchronization methods for video fingerprint have been presented, which are implemented in raw data. In this paper, a compression compliant video desynchronization method for collusion resilient fingerprint is proposed. The technique can simultaneously apply random space/time desynchronization and compression to videos. In our experiments, with little visual degradation, the joint process costs no more time and bandwidth than those of MPEG2 encoding/decoding. By this method, the video quality degrades dramatically for colluded copies. Besides evaluating the method by compression quality, compression time, and visual quality, we also discussed the system security. Two attacks are considered for the security evaluation: re-synchronization attack (including Most Similar Frame Collusion and Random Similar Frame Replacement) and re-desynchronization attack. Schemes for robustness to these attacks are also shown. Two theorems are presented to point out the security limit of single time desynchronization and single space desynchronization and the influence of related parameters to security.

This paper proposes a histogram shifting method for image lossless data hiding in integer wavelet transform domain. This algorithm hides data into wavelet coefficients of high frequency subbands. It shifts a part of the histogram of high frequency wavelet subbands and thus embeds data by using the created histogram zero-point. This shifting process may be sequentially carried out if necessary. Histogram modification technique is applied to prevent overflow and underflow. The performance of this proposed technique in terms of the data embedding payload versus the visual quality of marked images is compared with that of the existing lossless data hiding methods implemented in the spatial domain, integer cosine transform domain, and integer wavelet transform domain. The experimental results have demonstrated the superiority of the proposed method over the existing methods. That is, the proposed method has a larger embedding payload in the same visual quality (measured by PSNR (peak signal noise ratio)) or has a higher PSNR in the same payload.

In sensitive imagery, such as deep space exploration, military reconnaissance and medical diagnosis, traditional watermarks can be hardly found useful. The main reason is that the users are too worried about the loss of original information after the image being embedded with other data. Although early watermarking methods only distort the host signal imperceptibly, there is still some host information that may be permanently (irreversibly) lost. To avoid this disadvantage, some researchers (e.g. [1]-[3]) proposed the concept of reversible (lossless) watermark. Recently, more and more reversible watermarking methods have been proposed. However, the influence of [1]-[3] is obvious. In this paper, we focus on analyzing and comparing these three reversible watermarking methods. Our investigation covers several aspects including data hiding capacity, image quality, capacity resilience and control, computational complexity, security, and blind data extraction. Such analysis and comparison provide indispensable information for the design of new reversible watermarking techniques.

In this paper, we propose a reversible watermarking algorithm where an original image can be recovered from watermarked image data. Most watermarking algorithms cause degradation of image quality in original digital content in the process of embedding watermark. In the proposed algorithm, the original image can be obtained when the degradation is removed from the watermarked image after extracting watermark information. In the proposed algorithm, we utilize a peak point of image histogram and the location map and modify pixel values slightly to embed data. Because the peak point of image histogram and location map are employed in this algorithm, there is no need of extra information transmitted to receiving side. Also, because a slight modification on pixel values is conducted, highly imperceptibly images can be achieved. As locations of watermark embedding are identified using location map, amount of watermark data can dramatically increases through recursive embedding. Experimental results show that it can embed 5K to 130K bits of additional data.

We propose two efficient information hiding algorithms in the least significant bits of JPEG coefficients of images. Our algorithms embed information by modifying JPEG coefficients in such a way that the introduced distortion is minimized. We derive the expected value of the additional error due to distortion as a function of the message length and the probability distribution of the JPEG quantization errors of cover images. We have implemented our methods in Java and performed the extensive experiments with them. The experiments have shown that our theoretical predictions agree closely with the actual introduced distortions.

Digital watermarking is increasingly demanded for protecting or verifying the original image ownership. In this paper, we propose an improved differential energy watermarking using adaptive differential energy watermarking (ADEW) with cross binding wavelet tree (CBWT). The ADEW embeds a secret bit string which is obtained by the error correction code (ECC). Thus, the proposed method not only takes advantage of the ADEW’s error resilience but corrects a secret bit string by using ECC after error occurrence. Through experiments, we compare the proposed method with conventional DEW approaches and the proposed method shows the appropriateness for robust watermarking.

Distributing video contents via broadcasting network mechanisms has become a promising business opportunity for the entertainment industry. However, since content piracy is always a serious problem, broadcasted contents must be adequately protected. Rather than implementing sophisticate key-management schemes for access control, an animation broadcast system based on colorization techniques is proposed. In the proposed system, gray-level animation video sequences are delivered via broadcast mechanisms, such as multicast, to reduce the overhead in server processing and network bandwidth. Moreover, color seeds labeled with fingerprint codes are delivered to each client through low-bandwidth auxiliary connections and then used to generate high-quality full-color animations with slight differences between versions received by each client-side device. When a user illegally duplicates and distributes the received video, his identity can be easily found out by examining features extracted from the pirated video. The proposed scheme also shows good resistance to collusion attacks where two or more users cooperate to generate an illegal copy in expectation of getting rid of legal responsibility. The proposed scheme exhibits advantages in network bandwidth, system performance and content security.

In this paper, we propose an adaptive blind video watermarking method using video characteristics based on human visual system (HVS) in three-dimensional discrete cosine transform (3D-DCT) domain. In order to optimize the weight factors for watermarking, we classify the patterns of 3D-DCT cubes and the types of video segments by using the texture and motion information of 3D-DCT cubes. Then we embed an optimal watermark into the mid-range coefficients of 3D-DCT cubes by using the trained optimal weight factors. Experimental results show that the proposed method achieves better performance in terms of invisibility and robustness than the previous method under the various possible attacks such as MPEG compression, frame dropping, frame insertion and frame swapping to experimental videos.

In this paper, we propose an encryption algorithm to protect scalable video coding (SVC) bitstream and a method for conditional access control to consume the encrypted SVC bitstream. To design an encryption algorithm, we analyzed the encryption requirements to support scalability function in the scalable video and proposed an effective encryption method developed in the network abstraction layer (NAL). In addition, conditional access control scheme and key management scheme are proposed to consume the SVC bitstream protected with proposed method. Experiments were performed to verify the proposed method and results showed that proposed algorithms could provide an effective access control of scalable video as well as support a video encryption with scalability function.

This paper proposes a wavelet-based fragile watermarking scheme for secure image authentication. In the proposed scheme, the embedded watermark is generated using the discrete wavelet transform (DWT), and then the improved security watermark by scrambling encryption is embedded into the least significant bit (LSB) of the host image. The proposed algorithm not only possesses excellent tamper localization properties and greater security against many attacks, but also demonstrates a new useful feature that can indicate whether the modification made to the image is on the contents or the embedded watermark. If only the watermark is modified, the authenticity of the image is assured, instead of being declared as a counterfeit. Experimental results illustrate the effectiveness of our method.

Secure watermark embedding allows to securely embed a watermark into a piece of content at an untrusted user device without compromising the security of the watermark key, the watermark or the original. In this paper, we show how secure embedding can be achieved by using traditional watermarking schemes in conjunction with partial encryption techniques, which were primarily developed to facilitate fast encryption of media content. Based on this concept, we develop two new efficient secure embedding mechanisms, one for the MASK watermarking scheme operating on baseband audio and one for a spread spectrum watermarking scheme operating on MPEG-2 encoded video streams.

Robustness and security are the key issues in the development of image watermarking algorithm. A new rotation invariant security image watermarking algorithm based on steerable pyramid transform is proposed in this paper. The algorithm is characterized as follows: (1) the rotation invariance and robust watermarking are achieved concurrently on the same transform domain; (2) the rotation synchronization is obtained through template matching by using steerable pyramid transform, which satisfies the shiftability in orientation condition; (3) the watermarks are embedded into an oriented subband at angle

θ

, which can be interpolated with base filter kernels and used as a key in watermark detection to increase the security of watermark; (4) the watermark detector is designed based on the steerable vector HMM model. High robustness is observed against StirMark attacks and their joint attacks.

The pervasive distribution of digital images triggers an emergent need of authenticating degraded images by lossy compression and transmission. This paper proposes a robust content-based image authentication scheme for image transmissions over lossy channels. Content-based image authentication typically assesses authenticity based on a distance measure of feature differences between the testing image and its original. Commonly employed distance measures such as the Minkowski measures may not be adequate for content-based image authentication since they do not exploit statistical and spatial properties of the feature differences. This proposed error resilient scheme is based on a statistics- and spatiality-based measure (

SSM

) of feature differences. This measure is motivated by an observation that most malicious manipulations are localized whereas acceptable manipulations cause global distortions. Experimental results show that

SSM

is better than previous used measures in distinguishing malicious manipulations from acceptable ones, and the proposed

SSM

-based scheme is robust to transmission errors and other acceptable manipulations, and is sensitive malicious image modifications.