No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Watermarking digital image and video data. A state-of-the-art overview
Abstract
The authors begin by discussing the need for watermarking and the
requirements. They go on to discuss digital watermarking techniques
based on correlation and techniques that are not based on correlation
... D IGITAL watermarking is an essential part of information hiding research [1], [2], [3]. It involves the process of embedding data in the form of watermarks in carrier (host) signals, which are typically images, audio (speech) or video data [4], [3]. However, in the existing literature, the utilization of other signals has also been described, such as in 2D or 3D computer graphics [5], 2D vector maps [6], 3D printing models [7], text [5], network flows [8], databases [5], [9], various kinds of nonmedia data which can be subject to data mining process [10], or machine learning models, particularly neural networks [11], [12]. ...
... The overview presented by Langelaar et al. [4] focused on the explanation of some watermarking algorithms and their possible improvements, rather than on some specific applications of watermarking or addressing particular types of attacks. Although the paper was dedicated to images and video, the vast majority of approaches presented there were suited for images, and only a few were designed to deal with videos; although some of the described methods can be applied to both images and videos. ...
... They also mentioned the question of audio watermarking, but none of the algorithms described were designed particularly for audio. Langelaar et al. [4] described typical watermarking applications, including copyright protection, fingerprinting, copy protection, broadcast monitoring, data authentication (defined as checking data authenticity, informing whether the content was modified, and providing localization of changes), indexing, medical safety (e.g., to include patient data in medical images), and data hiding. The authors also explained the most important features of watermarking schemes, including transparency, payload, robustness, security, and blindness, and indicated the relationships between them. ...
During the past few decades, research on digital media watermarking –initially designed for digital images with the envisioned applications of copyright protection or copy control– has significantly evolved with respect to other covers (i.e., video, audio, speech) and many more potential applications, including tamper detection, broadcast monitoring, and, more recently, fake news detection. As a result, various surveys have tried to summarize certain aspects of this research field as it has grown. This has led to more than 130 survey papers being written at different points in time, describing various parts of the scientific efforts focused on digital media watermarking. Considering the above, the aim of this paper is twofold. First, we conduct a meta-survey based on 64 selected research works, in order to summarize the most notable survey papers in this field, which allows us to “draw a map” of this research area. Second, we focus on providing the requirements for digital watermarking techniques when applied to their most recent application: detecting fake news in multimedia content. Finally, an outline of the approach taken within the DISSIMILAR (Detection of fake newS on SocIal MedIa pLAtfoRms) project for the detection of disinformation is presented.
... Encryption offers the security of digital images in transit while watermarking imperceptibly watermark remains even after decryption. Image watermarking has traditionally been researched for purposes: copyright protection, fingerprinting, copy protection, broadcast monitoring, data authentication, indexing, medical safety, and data hiding [22,29,30]. This section contains an explanation regarding image watermarking areas/applications concerning medical images. ...
... When a medical image contains patient data such name, patient ID for prevention against false ownership such can be refer to as medical image watermarking for copyright protection. This type of watermarking application detects illegal manipulation but cannot protect against duplication [22,29,31]. Evaluation of existing research indicates the following [32,33] uses copyright protection for medical images. ...
... Ouazzane et al. described the Authentication and integrity verification as a reliability approach, a fragile watermarking with tamper detection, localization, and tamper recovery. This approach is sometimes used along with transform domain approach to presents a robust, imperceptible and authentication scheme [21,22,29]. ...
Medical images are fundamentally utilized for rendering diagnostic and treatment to patients. Medical images are a patient body or part captured using medical imaging devices such as CT, X-Ray, PET, MRI, and US. Technological advancement introduces E-Healthcare systems, Telemedicine, and Electronic Health Information Systems (EHIS) enabling medical images to flow over the public network for remote healthcare services. The manipulation or replacement of medical images is fatal to the well-being of a patient, thereby requiring protection using watermarking. Watermarking is a data security approach toward protecting medical images against abuse by unauthorized personnel via providing confidentiality, authentication, and integrity verification. The dynamism and importance of medical image watermarking require constant literature update on trends, issues, and challenges which leading to the forgoing research survey. The survey proposes to highlight trending application areas in medical image watermarking research and evaluation of the recent approach adopted by researchers. Furthermore, the survey evaluates existing work in compliance with the standard benchmark requirement in design and performance and presents a discussion on the way forward to other possible research opportunities in the medical image watermarking domain.
... Therefore, copyright protection of images is necessary when shared on open access. Digital watermarking is a technology that solves this problem [1][2][3][4][5][6][7][8][9][10][11]. In watermarking, user identity information in terms of the watermark is inserted into cover content using the gain factor to generate watermarked content. ...
... In watermarking, user identity information in terms of the watermark is inserted into cover content using the gain factor to generate watermarked content. Watermarking can classify in various ways, such as types of cover content, processing domain, nature against attacks, and types of watermark extraction methods [1,6]. Based on the types of cover content, watermarking can be classified as text watermarking, image watermarking, and signal watermarking. ...
... Two of these watermarking requirements parameters can be met without sacrificing the third, but if two are met, the third may suffer. In the literature [1,6], several researchers presented an imperceptible and robust scheme that compromised the scheme's payload capacity. Therefore, increasing the payload capacity of watermarking schemes is one of the open research areas. ...
The main issues in any watermarking scheme are the robustness and transparency of color images against different manipulations. The watermarked images are generated using a gain factor in many watermarking techniques. But the selection of gain factor is a significant issue in many watermarking techniques developed by manual or trial and error processes. Therefore, the optimization algorithm used for the generation of the optimal gain factor is used to generate watermarked images. The proposed method uses particle swarm optimization (PSO) to create an optimal gain factor. In contrast, a block redundant wavelet transform (RDWT) is used to generate color watermarked images. Here, RDWT is applied to the color cover image to get coefficients of different wavelet subbands. Noise sequences modify the coefficients of detail wavelet subbands of the color cover image in this scheme according to watermark bits. The experimental results and comparative analysis of the scheme with existing schemes show that the proposed scheme performs better in imperceptibility, robustness, and payload capacity.
... In this method the information is hidden by using linearly combining the cover image with a small pseudo noise signal which is modulated by embedded watermark [7]. Code-Division Multiple Access (CDMA) spread spectrum is a popular method and robust to cropping because the watermark bits are scattered in a random way over all the host image based on PN sequence which is generated using independent seeds for each watermark value [14,15]. Also this approach can be applied in frequency domain by using Discrete Wavelet Transform (DWT) or Discrete Cosine Transform (DCT). ...
... CDMA [16] CDMA [16,17] Scaling CDMA [16] CDMA [16,17] Cropping CDMA [14,15] CDMA [16,17] Sharpen DWT [23,26], DCT [21] Salt-and-pepper LSB [11,12,13] DWT [23,26] Gaussian LSB [11,12,13] DWT [23,26] Median filtering LSB [11,12,13] JPEG-Compression CDMA [16,17], DCT [21] JPEG2000 DWT [24] Histogram Equalization DWT [23,26] Overall, most of the spatial domain techniques tend to be less robust and secure compared to frequency domain techniques against different types of attacks, especially DWT. It has more advantages than other transformation, because DWT allows watermark to be embedded in regions where Human Visual System (HVS) is less sensitive. ...
Multimedia protection has become a significant problem in the society where the intellectual property is threatened. Digital watermarking can be considered as a solution to protect the multimedia. Property protection is a common field of research compared to other fields such as identity verification and local manipulation localization. Imperceptibility and robustness are the most crucial requirements for multimedia watermarking. High distortion is provided by watermarks with high robustness. We must so strike a balance between them. Recently many works based on frequency range which can satisfy watermarking requirements such as high robustness and low distortion. Visual attention based image watermarking is considered one of image watermarking methods due to detect the most important regions for watermark embedding. The lack of robustness against malicious cyber-attacks makes watermarks easily detectable and destroyed. As a result, the proposed watermarking scheme becomes more complex and cannot withstand diversity Geometric and non-geometric attacks. Therefore, there are many existing visual attentions based image watermarks. This paper analyzes the techniques for image watermarking against various types of attacks and presents the applications of digital watermarking.
... Because of this, if an image is shared on an open-access medium, it must be protected by copyright. Watermarking can be used to solve this problem [1][2][3][4][5][6][7][8][9][10][11]. Using an embedding factor, watermarked content is generated from the cover image. ...
... Using an embedding factor, watermarked content is generated from the cover image. There are many types of watermarking based on the content of their covers, the processing domains, the attack, and the extraction method [1,6]. Watermarks can be applied as text watermarks, image watermarks, or signal watermarks, depending on the content of the cover. ...
A medical image watermarking system must meet three basic requirements: transparency, robustness, and payload capacity. Watermarking medical images is mostly concerned with ensuring they are robust against manipulation. The embedding factor is used in many robust watermarking algorithms to hide the secret information behind the cover medical image. Many watermarking algorithms have difficulty selecting appropriate embedding factors. An effective copyright protection algorithm for medical images uses particle swarm optimization (PSO) to provide robust, hybrid, and blind watermarking. The embedding factor is selected using PSO based on the medical image and watermark information. Compared to the existing algorithms, the proposed algorithm performs better in transparency and payload capacity based on experimental results and comparative analysis.
... Digital watermarking allows us to embed hidden data into a host content (image, speech, music, video, etc) for a variety of important applications: copyright protection, fingerprinting, copy protection, broadcast monitoring, data authentication, indexing, data hiding and others applications [1], [2], [3]. An exponentially increasing number of publications has appeared on the subject during the last decade, indicating the increasing interest of the scientific community [2]. ...
... One of most interesting propoerties is that narrow band watermarks are spread over many frequency bins so that the energy in any given bin is very small and hardly detectable [4]. Many works use the concept of spread spectrum communications to understand watermark insertion as information transmission over a noisy channel [4], [1], where the noise in this case is the host content. The implementation of efficient spread spectrum watermarking requires the generation of a set of patterns with very specific properties. ...
... Because of this, if an image is shared on an open-access medium, it must be protected by copyright. Watermarking can be used to solve this problem [1][2][3][4][5][6][7][8][9][10][11]. Using an embedding factor, watermarked content is generated from the cover image. ...
... Using an embedding factor, watermarked content is generated from the cover image. There are many types of watermarking based on the content of their covers, the processing domains, the attack, and the extraction method [1,6]. Watermarks can be applied as text watermarks, image watermarks, or signal watermarks, depending on the content of the cover. ...
There are three basic requirements for medical image watermarking: transparency, robustness, and payload capacity. The main issues in medical image watermarking are the robustness of medical images against various manipulations. In many robust watermarking algorithms, the secret information is hidden behind the cover medical image using the embedding factor. But selecting appropriate embedding factors is an issue in many watermarking algorithms. This paper proposes a robust, hybrid, and blind watermarking algorithm with particle swarm optimization (PSO) for copyright protection of medical images. Here, PSO is used to select an appropriate embedding factor according to medical image and watermark information. The experimental results and comparative analysis of the algorithm with existing algorithms show that the proposed algorithm performs better in transparency and payload capacity.
... More precisely, we use the backdoor triggers as digital watermarks to identify the ownership of a GNN model. Indeed, digital watermarking is typically used to identify ownership of the copyright of media signals, e.g., audio, video, and image data [13]. There are already some works discussing embedding watermarks into DNN models to protect the IP of these models. ...
... Digital watermarking is a technique that embeds certain watermarks in carrier multimedia data such as audio, video, or images to protect their copyright [13]. The information to be embedded in a signal is called a digital watermark. ...
Graph Neural Networks (GNNs) have achieved promising performance in various real-world applications. Building a powerful GNN model is not a trivial task, as it requires a large amount of training data, powerful computing resources, and human expertise on fine-tuning the model. What is more, with the development of adversarial attacks, e.g., model stealing attacks, GNNs raise challenges to model authentication. To avoid copyright infringement on GNNs, it is necessary to verify the ownership of the GNN models. In this paper, we present a watermarking framework for GNNs for both graph and node classification tasks. We 1) design two strategies to generate watermarked data for the graph classification and one for the node classification task, 2) embed the watermark into the host model through training to obtain the watermarked GNN model, and 3) verify the ownership of the suspicious model in a black-box setting. The experiments show that our framework can verify the ownership of GNN models with a very high probability (around $100\%$) for both tasks. In addition, we experimentally show that our watermarking approach is still effective even when considering suspicious models obtained from different architectures than the owner's.
... Classifier-based methods are widely spread in detection paradigms, while watermarkbased methods represent an innovative alternative to the above-mentioned methods [52]. In the early days, watermark methods were used in computer vision and image processing to ensure copyright protection [53]. Recently, Kirchenbauer [54] proposed in their study the use of watermarks with LLMs, incorporating signals in generated text, which is undetectable to human observers and can be detected with open source algorithms without access to the language model API or parameters for detection. ...
Machine-generated content reshapes the landscape of digital information; hence, ensuring the authenticity of texts within digital libraries has become a paramount concern. This work introduces a corpus of approximately 60 k Romanian documents, including human-written samples as well as generated texts using six distinct Large Language Models (LLMs) and three different generation methods. Our robust experimental dataset covers five domains, namely books, news, legal, medical, and scientific publications. The exploratory text analysis revealed differences between human-authored and artificially generated texts, exposing the intricacies of lexical diversity and textual complexity. Since Romanian is a less-resourced language requiring dedicated detectors on which out-of-the-box solutions do not work, this paper introduces two techniques for discerning machine-generated texts. The first method leverages a Transformer-based model to categorize texts as human or machine-generated, while the second method extracts and examines linguistic features, such as identifying the top textual complexity indices via Kruskal–Wallis mean rank and computes burstiness, which are further fed into a machine-learning model leveraging an extreme gradient-boosting decision tree. The methods show competitive performance, with the first technique’s results outperforming the second one in two out of five domains, reaching an F1 score of 0.96. Our study also includes a text similarity analysis between human-authored and artificially generated texts, coupled with a SHAP analysis to understand which linguistic features contribute more to the classifier’s decision.
... This feature has led to the widespread use of digital watermarking in the field of multimedia copyright protection and authentication by inserting author logotypes as evidence in situations of copyright disputes. The technique has been extensively applied in protecting the property rights of images [5], [6], videos [7], [8], and audio ...
Deep learning-based image watermarking algorithms have been widely studied as an important technology for copyright protection. These methods utilize an end-to-end architecture with an encoder, a noise layer and a decoder to make the watermark robust to various distortions. However, recent algorithms present unsatisfactory visual quality and robustness against JPEG compression, which is the most common image processing operation but is non-differential thus cannot be directly included in the noise layer. To address this limitation, this study proposes a novel enhanced attention-based image watermarking algorithm with simulated JPEG compression, which leverages the channel and spatial attention mechanism to facilitate watermark embedding and simulates JPEG compression with a suitably designed function. Precisely, we design a differentiable rounding function based on the Fourier series to replace the quantization process in JPEG compression, which overcomes the non-differentiability of JPEG compression and can be incorporated in the training process. In addition, we propose an enhanced dual attention module in the encoder, which combines channel and spatial attention to improve the performance of our model. The channel attention guides the encoder to fuse the watermark into more important channels and the spatial attention further helps to embed the watermark into regions with more complex textures. The experimental results show that our method generates high quality watermarked images, with PSNR over 50 when no noise is applied. Compared with current methods, our model achieves stronger robustness to JPEG compression, with bit accuracy over 99% under the JPEG compression with quality factor of 50. Besides, the proposed framework also exhibits excellent robustness for a variety of common distortions, including cropout and dropout.
... В предлагаемом методе вычисленные коэффициенты ДКП для каждого патча (квадратный фрагмент изображения) делятся на три группы, а именно, на соответствующие низким, средним и высоким частотам [5]. Рисунок 3 иллюстрирует эти три группы коэффициентов на примере патча размером 7 × 7 пикселей. ...
The article is devoted to the problem of detecting blurred areas in high-resolution full-slide histological images. The proposed method is based on the use of two approaches: blur detection using multiscale analysis of discrete cosine transform coefficients and assessing the degree of sharpness of object boundaries in the image. The effectiveness of the algorithm is verified on synthetically blurred images as well as real full-slide images from the PATH-DT-MSU dataset.
... Even though, many-a-times videos are uploaded and streamed online without any protection [41,42]. Several video watermarking techniques are available to watermark the videos both in the uncompressed and compressed domains [43][44][45]. ...
The widespread expansion of internet and digital sharing of multimedia content has led to a huge rise in security concerns like ownership, content authentication and copyright protection of the multimedia data. This paper presents a novel scheme for watermarking of MPEG-4 videos based on the Extreme Learning Machine (ELM) for frame selection task and hybridization of Fuzzy logic with Particle Swarm Optimization technique for watermark embedding and extraction process. The ELM is applied on the host video to select the relevant frames fit for watermarking process. The embedding step is carried out in DWT-SVD domain using Fuzzy-PSO technique. The proposed scheme is tested over five standard videos having advanced video coding (AVC) format. The robustness and visual quality testing post embedding is respectively carried out by computing Normalized Cross-Correlation (NC(W,W’)) and Bit Error Rate (BER(W,W’)) along with Average Peak-Signal-to-Noise Ratio (\({PSNR}_{avg}\)) parameters by applying eight different attacks. Our proposed scheme is tested and compared for its outcomes with those of different other standard schemes. It is found that our results are superior to all other frontline schemes.
... Inspired by copyright protection watermarks in the image and video fields (Langelaar et al., 2000), some works (Meral et al., 2009;Krishna et al., 2023;Kirchenbauer et al., 2023) explore the potential of watermarks in language models, which modify model generation behaviors to make them easier to detect, becoming a new detection perceptiveness with magic shortcuts. Our work does not assume language models are enhanced with watermarks, instead considering a more common detection setting where we do not know the sources of detected texts. ...
Recent advances in large language models have enabled them to reach a level of text generation comparable to that of humans. These models show powerful capabilities across a wide range of content, including news article writing, story generation, and scientific writing. Such capability further narrows the gap between human-authored and machine-generated texts, highlighting the importance of deepfake text detection to avoid potential risks such as fake news propagation and plagiarism. However, previous work has been limited in that they testify methods on testbed of specific domains or certain language models. In practical scenarios, the detector faces texts from various domains or LLMs without knowing their sources. To this end, we build a wild testbed by gathering texts from various human writings and deepfake texts generated by different LLMs. Human annotators are only slightly better than random guessing at identifying machine-generated texts. Empirical results on automatic detection methods further showcase the challenges of deepfake text detection in a wild testbed. In addition, out-of-distribution poses a greater challenge for a detector to be employed in realistic application scenarios. We release our resources at https://github.com/yafuly/DeepfakeTextDetect.
... One of the most exciting works in recent times around this research revolves around watermarking and developing efficient watermarks for machine-generated text detection. Historically, watermarks have been employed in the realm of image processing and computer vision to safeguard copyrighted content and prevent intellectual property theft (Langelaar et al., 2000). They can also be used for data hiding, where information is hidden within the watermark itself, allowing for secure and discreet transmission of information. ...
Our work focuses on the challenge of detecting outputs generated by Large Language Models (LLMs) from those generated by humans. The ability to distinguish between the two is of utmost importance in numerous applications. However, the possibility and impossibility of such discernment have been subjects of debate within the community. Therefore, a central question is whether we can detect AI-generated text and, if so, when. In this work, we provide evidence that it should almost always be possible to detect the AI-generated text unless the distributions of human and machine generated texts are exactly the same over the entire support. This observation follows from the standard results in information theory and relies on the fact that if the machine text is becoming more like a human, we need more samples to detect it. We derive a precise sample complexity bound of AI-generated text detection, which tells how many samples are needed to detect. This gives rise to additional challenges of designing more complicated detectors that take in n samples to detect than just one, which is the scope of future research on this topic. Our empirical evaluations support our claim about the existence of better detectors demonstrating that AI-Generated text detection should be achievable in the majority of scenarios. Our results emphasize the importance of continued research in this area
... If visible information is embedded in the media as a watermark, the watermark is termed as visible digital watermark. This may be a logo or a text that marks a digital medium (Po-Chyi et al., 2017;Langelaar et al., 2000). ...
A digital watermark, which is embedded in an image sequence or video frames as the form of a binary string or visual logo, is a small size of visible data. Thus, the quality of embedded videos is often slashed due to the watermarking. Comparative video watermarking is a highly innovative method that was designed to unravel this issue. In this chapter, the authors make use of singular value decomposition (SVD) and discrete wavelet transform (DWT) for video watermarking; the authors employed inverse transform (IDWT) to extract the video watermark. The digital signature is also utilised to increase the authenticity of watermarks and verify any changes. The authors combine this approach with digital fingerprinting as well as to get the improved results. Throughout the designed attacks, the merits of the new watermarking paradigm such as robustness, convergence, and stability are attained with security and authenticity by calculating the metrics such as MSE, PSNR, entropy, SSIM, etc.
... In the proposed work, the watermarking system is designed significantly involving all the above applications. The watermarking systems are categories into two types, they are spatial and frequency domain watermarking, and the tradeoff parameters are robustness, imperceptibility, payload, and security [8]. In the literature, it is a challenge to propose a spatial domain watermarking system by maintaining the trade-off parameters. ...
Digital raw images obtained from the data set of various organizations require authentication, copyright protection, and security with simple processing. New Euclidean space point’s algorithm is proposed to authenticate the images by embedding binary logos in the digital images in the spatial domain. Diffie–Hellman key exchange protocol is implemented along with the Euclidean space axioms to maintain security for the proposed work. The proposed watermarking methodology is tested on the standard set of raw grayscale and RGB color images. The watermarked images are sent in the email, WhatsApp, and Facebook and analyzed. Standard watermarking attacks are also applied to the watermarked images and analyzed. The finding shows that there are no image distortions in the communication medium of email and WhatsApp. But in the Facebook platform, raw images experience compression and observed exponential noise on the digital images. The authentication and copyright protection is tested from the processed Facebook images, it is found that the embedded logo could be recovered and seen with added noise distortions. So the proposed method offers authentication and security with compression attacks. Similarly, it is found that the proposed methodology is robust to JPEG compression, image tampering attacks like collage attack, image cropping, rotation, salt, and pepper noise, sharpening filter, semi-robust to Gaussian filtering, and image resizing, and fragile to other geometrical attacks. The receiver operating characteristics (ROC) curve is drawn and found that the area under the curve is approximately equal to unity and restoration accuracy of [67 to 100]% for various attacks.
... Visual quality is related to the difference between the original image with respect to the image after data hiding whereas the data hiding capacity is the capacity of hiding maximum data in the cover image pixels. There are three domains namely, spatial domain, compression domain, and frequency domain, in which the data can be concealed [17]. In the case of the spatial domain, the intensity values of pixels are altered to innocuously hide the secret information. ...
Communication bandwidth plays a significant role in real-time communication. For this, absolute moment block truncation coding (AMBTC) has been popular. An AMBTC based high capacity multimedia data hiding method for covert communication is proposed in this paper. The proposed method applies AMBTC to the host image to get the compressed image in the form of AMBTC trios. The trios are then classified into three categories based on the difference between their quantization levels. The proposed method then adaptively conceals the secret message bits in the bitmaps of the trios based on their type. Additionally, the proposed method conceals two bits of secret data in every combination of quantization levels irrespective of their block types using the LSB substitution technique. The experimental results show that the proposed AMBTC based method is perform better to the other related data hiding based methods in terms of embedding capacity while providing comparable PSNR.
... In the proposed work, the watermarking system is designed significantly involving all the above applications. The watermarking systems are categories into two types, they are spatial and frequency domain watermarking, and the trade-off parameters are robustness, imperceptibility, payload, and security [8]. In the literature, it is a challenge to Abstract Digital raw images obtained from the data set of various organizations require authentication, copyright protection, and security with simple processing. ...
Digital raw images obtained from the data set of various organizations require authentication, copyright protection, and security with simple processing. New Euclidean space point’s algorithm is proposed to authenticate the images by embedding binary logos in the digital images in the spatial domain. Diffie–Hellman key exchange protocol is implemented along with the Euclidean space axioms to maintain security for the proposed work. The proposed watermarking methodology is tested on the standard set of raw grayscale and RGB color images. The watermarked images are sent in the email, WhatsApp, and Facebook and analyzed. Standard watermarking attacks are also applied to the watermarked images and analyzed. The finding shows that there are no image distortions in the communication medium of email and WhatsApp. But in the Facebook platform, raw images experience compression and observed exponential noise on the digital images. The authentication and copyright protection are tested from the processed Facebook images. It is found that the embedded logo could be recovered and seen with added noise distortions. So the proposed method offers authentication and security with compression attacks. Similarly, it is found that the proposed methodology is robust to JPEG compression, image tampering attacks like collage attack, image cropping, rotation, salt-and-pepper noise, sharpening filter, semi-robust to Gaussian filtering, and image resizing, and fragile to other geometrical attacks. The receiver operating characteristics (ROC) curve is drawn and found that the area under the curve is approximately equal to unity and restoration accuracy of [67 to 100]% for various attacks.
... Digital watermarking technology [14,15] has powerful anticounterfeiting and antitheft capabilities and has been immensely leveraged to protect the IP of multimedia content. Motivated by such an intuition, DNN watermarking [16] has been proposed to protect the IP of DNNs. ...
Deep neural networks (DNN) with incomparably advanced performance have been extensively applied in diverse fields (e.g., image recognition, natural language processing, and speech recognition). Training a high-performance DNN model requires a lot of training data and intellectual and computing resources, which bring a high cost to the model owners. Therefore, illegal model abuse (model theft, derivation, resale or redistribution, etc.) seriously infringes model owners’ legitimate rights and interests. Watermarking is considered the main topic of DNN ownership protection. However, almost all existing watermarking works apply solely to image data. They do not trace the unique infringing model, and the adversary easily detects these ownership verification samples (trigger set) simultaneously. This paper introduces TADW, a dynamic watermarking scheme with tracking and antidetection abilities in the deep learning (DL) textual domain. Specifically, we propose a new approach to construct trigger set samples for antidetection and innovatively design a mapping algorithm that assigns a unique serial number (SN) to every watermarked model. Furthermore, we implement and detailedly evaluate TADW on 2 benchmark datasets and 3 popular DNNs. Experiment results show that TADW can successfully verify the ownership of the target model at a less than 0.5% accuracy cost and identify unique infringing models. In addition, TADW is excellently robust against different model modifications and can serve numerous users.
... Robustness is the resistance of an embedded watermark against intentional attack and normal signal processing operations such as noise, filtering, rotation, scaling, cropping and lossy compression etc. Capacity is the amount of data can be represented by embedded watermark. Many digital watermarking schemes have been proposed for still images and videos [2]. Most of them operate on uncompressed videos [3][4], while others embed watermarks directly into compressed videos [3,6]. ...
Watermarking techniques are mainly used for protecting intellectual property right. This paper proposes a new hybrid nonblind video watermarking technique using wavelet contourlet transform and nonnegative matrix factorization Wavelet transform processed images are losing edge information. The Contourlet transform has good approximation properties for smooth 2D functions and finds a direct discrete space construction. But its performance is considered to be redundant. There evolved wavelet based contourlet transform (WBCT), as a nonredundant version of the contourlet transform. WBCT is used for watermarking video frames. The non negative matrix factorization (NMF) is used as dimension reduction technique in watermarking. NMF is applied to low pass and directional high pass sub bands which results from WBCT of each original video frame and gray scale watermark images. Embedding action is performed in low pass sub-band of WBCT processed video frame. The hybrid scheme improves the performance of watermarking scheme. The experimental results shows that the proposed video watermarking scheme provides better video processing operations such as cropping, rotation, histogram equalization ,compression, variety of noises , frame dropping, frame averaging and frame swapping and etc.
... The semi-fragile watermark can survive in general image processing operations and it is also susceptive to malicious attacks. Due to this good property, semi-fragile watermarking has attracted great attention in image authentication and recovery [4]. Using fragile watermarking the most typical one for images is the watermarking algorithm based on the least significant bit (LSB). ...
... Digital watermarking involves the insertion of a certain code called as digest into the image right at its acquisition time. This is later used for an image authentication process which includes comparing extracted digest with the original digest [98][99][100]. If this extracted digest and original digest do not match then it means that some modification has happened to the image. ...
With the advent of Internet, images and videos are the most vulnerable media that can be exploited by criminals to manipulate for hiding the evidence of the crime. This is now easier with the advent of powerful and easily available manipulation tools over the Internet and thus poses a huge threat to the authenticity of images and videos. There is no guarantee that the evidences in the form of images and videos are from an authentic source and also without manipulation and hence cannot be considered as strong evidence in the court of law. Also, it is difficult to detect such forgeries with the conventional forgery detection tools. Although many researchers have proposed advance forensic tools, to detect forgeries done using various manipulation tools, there has always been a race between researchers to develop more efficient forgery detection tools and the forgers to come up with more powerful manipulation techniques. Thus, it is a challenging task for researchers to develop h a generic tool to detect different types of forgeries efficiently. This paper provides the detailed, comprehensive and systematic survey of current trends in the field of image and video forensics, the applications of image/video forensics and the existing datasets. With an in-depth literature review and comparative study, the survey also provides the future directions for researchers, pointing out the challenges in the field of image and video forensics, which are the focus of attention in the future, thus providing ideas for researchers to conduct future research.
... In addition, DWT is growing in popularity given its ability to facilitate transformation, which greatly strengthens the security of watermarked digital images. The DWT domain divides the image into several resolution levels and a starting set of processes ranging from highest to lowest resolution [16]. The measure of durability is thus increased by masking watermarks using greater intensity during digital images. ...
The propagation of digital media over the Internet has helped improve digitization, which has given an excessive lead to copyright issues. Digital watermarking techniques have been applied to address copyright issues. In research, a system is being developed to handle veracious types of watermarked attacks, for obtaining extreme security and an adequate level of visibility and robustness. The discrete wave transform (DWT) and singular value decomposition (SVD) approaches were applied to analyze veracious types of attacks. The DWT method was used to embed the host image in four levels; this level was processed using the SVD method. The peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) were applied to measure the invisibility, and the normalization correlation (NC) was used to examine the robustness of watermarked images. The empirical results showed that the proposed DWT-SVD achieved superior accuracy in identifying the various attacks. The proposed DWT-SVD performance was confirmed during the training process, and the proposed system was shown to have high invisibility and robustness against various types of attacks on watermarked images. Finally, the results of the proposed system were compared to existing systems, and it was shown that DWT-SVD achieved better performance in terms of pixel-value modification attacks.
... ese techniques are usually split into two types, referred to as active and passive detection techniques [7][8][9]. In active detection, a message digest or digital signature [10][11][12][13][14] is injected inside an image when it is created. In this forgery detection technique, statistical information such as mean, median, and mode is inserted into an image using some encryption method; this information is then retrieved from the image at the receiving side using a decryption method to check its authenticity [15]. ...
With the technological advancements of the modern era, the easy availability of image editing tools has dramatically minimized the costs, expense, and expertise needed to exploit and perpetuate persuasive visual tampering. With the aid of reputable online platforms such as Facebook, Twitter, and Instagram, manipulated images are distributed worldwide. Users of online platforms may be unaware of the existence and spread of forged images. Such images have a significant impact on society and have the potential to mislead decision-making processes in areas like health care, sports, crime investigation, and so on. In addition, altered images can be used to propagate misleading information which interferes with democratic processes (e.g., elections and government legislation) and crisis situations (e.g., pandemics and natural disasters). Therefore, there is a pressing need for effective methods for the detection and identification of forgeries. Various techniques are currently employed for the identification and detection of these forgeries. Traditional techniques depend on handcrafted or shallow-learning features. In traditional techniques, selecting features from images can be a challenging task, as the researcher has to decide which features are important and which are not. Also, if the number of features to be extracted is quite large, feature extraction using these techniques can become time-consuming and tedious. Deep learning networks have recently shown remarkable performance in extracting complicated statistical characteristics from large input size data, and these techniques efficiently learn underlying hierarchical representations. However, the deep learning networks for handling these forgeries are expensive in terms of the high number of parameters, storage, and computational cost. This research work presents Mask R-CNN with MobileNet, a lightweight model, to detect and identify copy move and image splicing forgeries. We have performed a comparative analysis of the proposed work with ResNet-101 on seven different standard datasets. Our lightweight model outperforms on COVERAGE and MICCF2000 datasets for copy move and on COLUMBIA dataset for image splicing. This research work also provides a forged percentage score for a region in an image.
1. Introduction
Digital images are used in almost every domain, such as public health services, political blogs, social media platforms, judicial inquiries, education systems, armed forces, businesses, and so on. Rapid advances in digital technology have led to the creation and circulation of a vast amount of images over the last few years. With the use of image/photo editing tools like Canva, CorelDRAW, PicMonkey, PaintShop Pro, and many other applications, it has become very easy to manipulate images and videos. Such digitally altered images are a primary source for spreading misleading information, impacting individuals and society. The deliberate manipulation of reality through visual communication with the aim of causing harm, stress, and disruption is a significant risk to society, given the increasing pace at which information is shared through social media platforms such as Twitter, Quora, and Facebook. It becomes a significant challenge for such social media platforms to identify the authenticity of these images. For example, cybersecurity experts [1] have reported that hackers have the ability to access patient’s 3-D medical scans and can edit or delete images of cancerous cells. In a recent study, surgeons were misled by scans modified with AI software, possibly leading to a high risk of misdiagnosis and insurance fraud. In addition, manipulated images related to politics [2] distributed across social media platforms have the potential to mislead and influence public perceptions and decisions. For example, studies have shown that that particular types of images are likely to be reused and, in certain cases, exploited in online terrorism communication channels through media sources [3–5]. Image alteration becomes too easy using image editing software and even altering the original image in such a way that forensic investigators will not be able to identify the changes in the image. The major camera manufacturers use digital certificates to solve this issue. However, some companies have generated forged images taken from Canon and Nikon camera models. These fake images are passed through manufacturer verification software to perform their authenticity test [6].
Therefore, there is a need to develop a forgery detection technique that detects and identifies forgeries to resolve these challenges. Many forgery detection techniques shown in Figure 1 have been developed to authorize a digital image. These techniques are usually split into two types, referred to as active and passive detection techniques [7–9]. In active detection, a message digest or digital signature [10–14] is injected inside an image when it is created. In this forgery detection technique, statistical information such as mean, median, and mode is inserted into an image using some encryption method; this information is then retrieved from the image at the receiving side using a decryption method to check its authenticity [15]. In passive detection, changes in the entire image and local features are identified. It does not leave any visual clues of forgery, but it alters the statistical information of an image. It verifies the structure and content of an image to determine its validity.
... The simplicity of implementation and the low complexity are the advantages of spatial-based watermarking schemes. On the other hand, they are not robust against common methods of attack (Wolfgang & Delp, 1998;Voyatzis & Pitas, 1998;Langelaar, Setyawan & Lagendijk, 2000;Hernández, Perez-Gonzalez & Rodriguez, 1998). To improve the spatial domain weakness, many methods are proposed. ...
The security of patient information is important during the transfer of medical data. A hybrid spatial domain watermarking algorithm that includes encryption, integrity protection and steganography is proposed to strengthen the information originality based on the authentication. The proposed algorithm checks whether the information of patients has been deliberately changed or not. The created code is distributed at every pixel of the medical image and not only in the regions of noninterest pixels, while the image details are still preserved. To enhance the security of the watermarking code, "SHA-1" is used to get the initial key for the Symmetric Encryption Algorithm. The target of this approach is to preserve the content of the image and the watermark simultaneously, this is achieved by synthesizing an encrypted watermark from one of the components of the original image and not by embedding a watermark in the image. To evaluate the proposed code the Least Significant Bit (LSB), Bit2SB, and Bit3SB were used. The evaluation of the proposed code showed that the LSB is of better quality but overall the Bit2SB is better in its ability against the active attacks up to a size of 2*2 pixels, and it preserves the high image quality.
Over the past few years, deep generative models have significantly evolved, enabling the synthesis of realistic content and also bringing security concerns of illegal misuse. Therefore, active protection for generative models has been proposed recently, aiming to generate samples with hidden messages for future identification while preserving the original generating performance. However, existing active protection methods are specifically designed for generative adversarial networks (GANs), restricted to handling unconditional image generation. We observe that they get limited identification performance and visual quality when handling audio-driven video generation conditioned on target audio and source input to drive video generation with consistent context, e.g. , identity and movement, between frame sequences. To address this issue, we introduce a simple yet effective active P rotection framework for A udio- D riven V ideo G eneration, named PADVG. To be specific, we present a novel frame-shared embedding module in which messages to hide are first transformed into frame-shared message coefficients. Then, these coefficients are assembled with the intermediate feature maps of video generators at multiple feature levels to generate the embedded video frames. Besides, PADVG further considers two visual consistent losses: i) intra-frame loss is utilized to keep the visual consistency with different hidden messages; ii) inter-frame loss is used to preserve the visual consistency across different video frames. Moreover, we also propose an auxiliary denoising training strategy through perturbing the assembled features by learnable pixel-level noise to improve identification performance, while enhancing robustness against real-world disturbances. Extensive experiments demonstrate that our proposed PADVG for audio-driven video generation can effectively identify the generated videos and achieve high visual quality.
In the rapidly evolving digital landscape, the generation of fake visual, audio, and textual content poses a significant threat to society’s trust, political stability, and integrity of information. The generation process has been enhanced and simplified using Artificial Intelligence techniques, which have been termed
deepfake
. Although significant attention has been paid to visual and audio deepfakes, there is also a burgeoning need to consider text-based deepfakes. Due to advancements in natural language processing and large language models, the potential of manipulating textual content to reshape online discourse and misinformation has increased. This study comprehensively examines the multifaceted nature and impacts of deep-fake-generated media. This work explains the broad implications of deepfakes in social, political, economic, and technological domains. State-of-the-art detection methodologies for all types of deepfake are critically reviewed, highlighting the need for unified, real-time, adaptable, and generalised solutions. As the challenges posed by deepfakes intensify, this study underscores the importance of a holistic approach that intertwines technical solutions with public awareness and legislative action. By providing a comprehensive overview and establishing a framework for future exploration, this study seeks to assist researchers, policymakers, and practitioners in navigating the complexities of deepfake phenomena.
Reversible data hiding in encrypted images (RDHEI) has gained significant popularity among security and privacy researchers as well as users, because of its features such as reversibility, embedding capacity (EC), and security. To enlarge the EC while ensuring the complete reversibility and security, we propose a bit-plane based RDHEI method based on multi-level blocking with quad-tree. The proposed method uses median edge detector (MED) as well as difference predictor to transform the original input image into a low-magnitude difference matrix. The difference matrix is then encoded by first employing a novel quad-tree based bit-plane representation strategy to exploit the intra-bit plane correlation and subsequently by inter bit-plane redundancy mitigation strategy to exploit inter bit-plane level correlation, for significantly condensing their size. Thus, a bigger room is reserved inside the cover image for embedding, so that a large amount of secret data can be hidden while ensuring the complete reversibility of the image. Experimental results validate the superiority of the proposed method over the state-of-the-art methods.
Image watermarking preserves digital content. This study introduces a new watermarking approach employing Sub-Band Discrete Cosine Transform and Deep neural networks, GRNN and CNN. The method embeds robust, invisible watermarks in greyscale photos and compares the two neural network topologies. The watermark is added using sub-band DCT. Watermark embedding in low-frequency sub-bands resists photo processing. The binary watermark modifies sub-band DCT coefficients to determine embedding intensity, resisting signal deterioration, and assaults. GRNN and CNN neural networks extract watermarks accurately. CNN extracts hierarchical features from images, enabling robust watermark recovery even under distortions, whereas non-parametric GRNN stores the whole training data to create predictions. The watermarking approach is tested on several greyscale photos. PSNR, SSIM, MSE, and NCC measure performance. The watermark tests noise addition, compression, and filtering. Compare GRNN and CNN's watermark extraction strengths and shortcomings to assess image watermarking suitability.
Watermarking the outputs of generative models is a crucial technique for tracing copyright and preventing potential harm from AI-generated content. In this paper, we introduce a novel technique called Tree-Ring Watermarking that robustly fingerprints diffusion model outputs. Unlike existing methods that perform post-hoc modifications to images after sampling, Tree-Ring Watermarking subtly influences the entire sampling process, resulting in a model fingerprint that is invisible to humans. The watermark embeds a pattern into the initial noise vector used for sampling. These patterns are structured in Fourier space so that they are invariant to convolutions, crops, dilations, flips, and rotations. After image generation, the watermark signal is detected by inverting the diffusion process to retrieve the noise vector, which is then checked for the embedded signal. We demonstrate that this technique can be easily applied to arbitrary diffusion models, including text-conditioned Stable Diffusion, as a plug-in with negligible loss in FID. Our watermark is semantically hidden in the image space and is far more robust than watermarking alternatives that are currently deployed. Code is available at github.com/YuxinWenRick/tree-ring-watermark.
Background
Steganography is the approach of camouflaging the covert object within another cover object. This manuscript suggested a novel steganography approach to conceal the covert data presence. The basic idea behind this is to generate an information-hiding approach that increases the payload capacity and good PSNR value without sacrificing much distortion of the image.
Objectives
To develop a novel data-hiding approach that increases imperceptibility, robustness, and payload capacity.
Methods
The Neighbour Mean Interpolation technique is used to scale up the original image to generate Interpolated pixels of the given image. An even-odd scheme on the interpolated stego pixel is used to camouflage the obscure code. MATLAB is used for the implementation of the new approach and results calculation.
Results
The Experimental analysis reveals that our suggested approach has a finer embedding capacity for camouflaging the secret data as the original image of size (MxN) is scaled up to size (2M-1 x2N-1) and also manages the good visuality of the cover or graven image. The proposed method is compared with Jung and Yoo, and Selvrani's method. The result of this comparison shows that the proposed method has finer imperceptibility than these two previously existing techniques.
Conclusion
A novel approach towards image steganography using neighbor mean interpolation has been proposed and implemented. A new steganography method is used for camouflaging the confidential code into the cover object using NMI without producing any major differences in the input image. The new approach provides better imperceptibility, robustness, and payload capacity.
Los grandes avances de la tecnología hoy en día han ayudado a la comunicación y distribución de todo tipo de material digital de manera instantánea. Este es uno de los grandes beneficios de las tecnologías de la información y comunicación, donde también existen diversas amenazas de infracciones relacionadas al derecho de autor o Copyright, falsificaciones de contenido, plagios, por mencionar algunas. Al pasar de los años se han desarrollado diferentes algoritmos computacionales para coadyuvar a la protección de datos en archivos de textos, audio y video, respectivamente. En este artículo se presenta un análisis del estado del arte sobre el uso de autómatas celulares dentro del campo de procesamiento de imágenes digitales en conjunto con la técnica de Marcado de Agua Digital (Digital Watermarking).
E-health care is an emerging field where health services and information are delivered and offered over the Internet. So the health information of the patients communicated over the Internet has to protect the privacy of the patients. The patient information is embedded into the health record and communicated online which also induces degradation to the original information. So, in this article, a zero watermarking scheme for privacy protection is proposed which protects the privacy and also eliminates the degradation done during embedding of patient information into the health record. This method is based on simple linear iterative clustering (SLIC) superpixels and partial pivoting lower triangular upper triangular (PPLU) factorization. The novelty of this article is that the use of SLIC superpixels and PPLU decomposition for the privacy protection of medical images (MI). The original image is subjected to SLIC segmentation and non-overlapping high entropy blocks are selected. On the selected blocks discrete wavelet transform (DWT) is applied and those blocks undergo PPLU factorization to get three matrices, L, U and P, which are lower triangular, upper triangular and permutation matrix respectively. The product matrix L×U is used to construct a zero-watermark. The technique has been experimented on the UCID, BOWS and SIPI databases. The test results demonstrate that this work shows high robustness which is measured using normalized correlation (NC) and bit error rate (BER) against the listed attacks.
Trigger set-based watermarking schemes have gained emerging attention as they provide a means to prove ownership for deep neural network model owners. In this paper, we argue that state-of-the-art trigger set-based watermarking algorithms do not achieve their designed goal of proving ownership. We posit that this impaired capability stems from two common experimental flaws that the existing research practice has committed when evaluating the robustness of watermarking algorithms: (1) incomplete adversarial evaluation and (2) overlooked adaptive attacks. We conduct a comprehensive adversarial evaluation of 11 representative watermarking schemes against six of the existing attacks and demonstrate that each of these watermarking schemes lacks robustness against at least two non-adaptive attacks. We also propose novel adaptive attacks that harness the adversary's knowledge of the underlying watermarking algorithm of a target model. We demonstrate that the proposed attacks effectively break all of the 11 watermarking schemes, consequently allowing adversaries to obscure the ownership of any watermarked model. We encourage follow-up studies to consider our guidelines when evaluating the robustness of their watermarking schemes via conducting comprehensive adversarial evaluation that includes our adaptive attacks to demonstrate a meaningful upper bound of watermark robustness.
This paper discussed a novel robust multiple watermarking method with three transform domain techniques, discrete wavelet transforms (DWT), discrete cosine transforms (DCT) and singular value decomposition (SVD) for color images. For security reasons, secret media is inserted into same multimedia items to offers an additional level of security and to achieve two important performance metrices of watermarking. Firstly, the original media (image) is disintegrated into first level DWT and generate DWT coefficients then select the lower frequency sub-band for applying DCT then SVD. The secret image is also altered by DCT decomposition then SVD transformation is applied on DCT coefficients in embedding process. Then after, a watermarked image is produced by applying converse of all transform domain SVD then DCT and DWT. The secret media can be extracted with recovery algorithm. This procedure has been generously tried and evaluated against various attacks of watermarking and it is found that it achieves better robustness and imperceptibility.
Son yıllarda internet kullanımında hem kullanıcı sayısının hem de bağlantı hızlarının
artması nedeniyle sayısal bilgilerin çoğaltılması ve işlenmesi yaygın hale gelmiştir. Bu
nedenle dijital resimlerin kopyalanma riskine karşı korunması ihtiyacı doğmuştur. Telif
hakkının korunması için önerilen yöntemlerden birisi de sayısal görüntünün
damgalanmasıdır. Bu çalışmada, ayrık kosinüs dönüşümü (AKD) uzayında katsayıların
değiştirilmesiyle dayanıklı bir sayısal damgalama yöntemi gerçekleştirmek ve mevcut
yöntemlerin başarımını arttırmak amaçlanmıştır. Çoklu-adaptif ölçekleme faktörleri
kullanan önerilen yöntemde, damgalanacak görüntünün tüm blokları uzaysal resim
frekansı (URF) metriği ile değerlendirilmiş ve her bloğa doku miktarıyla orantılı olarak
bir ölçekleme faktörü atanmıştır. Böylece, damgalanan görüntünün en az bozulmayla
ataklara karşı en fazla dayanımı göstermesi amaçlanmıştır. Gerçekleştirilen deneylerde,
önerilen çoklu-adaptif ölçekleme faktörü kullanan damgalama yönteminin sabit ve ikiliadaptif ölçekleme faktörü kullanan yöntemlere göre aynı saydamlık düzeyinde ataklara
karşı daha başarılı olduğu görülmüştür
Video watermarking technology has attracted increasing attentions in the past few years, and a great deal traditional and deep learning-based methods have been proposed. However, these existing methods usually suffer from the following two challenges: First, most algorithms cannot resist camcorder recording attack, which limits their practical application. Second, watermark embedding may cause substantial degradation of video quality. Through analyzing the unique distortions presented in the camcorder recording process, including geometric distortion, temporal sampling distortion, sensor distortion and processing distortion, this paper proposes a novel spatio-temporal context based adaptive camcorder recording watermarking scheme STACR. In STACR, considering the geometric distortion and video visual quality, we embed the watermark by constructing a spatio-temporal histogram and incorporate a content features based adaptive locating algorithm to select embedding blocks and embedding strengths. As for the temporal sampling attack, we put forward a watermark correlation-based synchronization algorithm and combine it with cross-validation. Moreover, to resist the sensor distortion, we design a local matching-based algorithm to improve the extraction accuracy. In addition, grouped and repeated embedding strategies are combined to cope with the processing distortion. Experimental results compared with the state-of-the-art show that the proposed scheme achieves high video quality and is robust to geometric attacks, compression, scaling, transcoding, recoding, frame rate changes and especially for camcorder recording.
Digital TV broadcasting needs new cryptological tools for
conditional access, copyright protection and image authentication. The
aim of this paper is to overview the corresponding systems' features.
The description of a conditional access system is given. It is shown
that equitable systems need the use of a trusted third party. The design
of efficient copyright protection by watermarking images and image
authentication by signatures are also briefly discussed
A w atermark is a perceptually unobstructive mark embed-ded in an image, audio or video clip or other multimedia asset. A w atermark can carry additional information, for in-stance about the source and copyright status of a document or its intended recipient, its rights and restrictions. We anal-yse the reliability of detecting such w atermarks, modeling it as a detection problem where the original content acts as noise or interference. Probabilities of incorrect detections are expressed in terms of the watermark-to-image power ra-tio, showing a signicant similarity in the problem of detect-ing watermarks and that of receiving weak spread-spectrum signals over a radio channel with strong interference. Theo-retical results are veried by experiments.
A new watermarking method is presented for still images and video streams. The method is different from nearly all known methods for image watermarking, which are based on adding pseudo-random noise to luminance or color components of the pixels. The new method is based on biasing the geometric locations of salient points in an image. The watermark is formed by a pre-defined dense pixel pattern, such as a collection of lines. So-called 'salient points' in the image are then modified, e.g. by warping or by changing the local luminance pattern around a salient point, such that after watermarking a majority of the new salient points lies on the watermark pattern. This paper describes the details of the new watermarking method and discusses the results of a series of tests performed on watermarked images. The feasibility and robustness of the method are shown.
In this paper we present a new watermarking technique for digital images. Our approach modifies blocks of the image after projecting them onto certain directions. By quantizing the projected blocks to even and odd values we can represent the hidden information properly. The proposed algorithm does the modification progressively to ensure successful data extraction without any prior information being sent to the receiver side. In order to increase the robustness of our watermark to scaling and rotation attacks we also present a solution to recover the original size and orientation based on a training sequence which is inserted as part of the watermark.
A textbook designed to guide students through the theory and practice of digital image processing. Examples are mainly satellite imagery (particularly Landsat TM data of Copenhagen), but medical and astronomical images are also included. Each chapter is balanced between the theory of the approach and its applications. Topics include: image display, filtering, Fourier transform, segmentation, geometric operations, and classifications. A mathematical background and some basic software written in PL/I, Pascal and Fortran are also included.-R.Harris image
Digital video data can be copied repeatedly without loss of quality. Therefore, copyright protection of video data is a more important issue in digital video delivery networks than it was with analog TV broadcast. One method of copyright protection is the addition of a "watermark" to the video signal which carries information about sender and receiver of the delivered video. Therefore, watermarking enables identification and tracing of different copies of video data. Applications are video distribution over the World Wide Web (WWW), pay-per-view video broadcast, or labeling of video discs and video tapes. In the mentioned applications, the video data is usually stored in compressed format. Thus, the watermark must be embedded in the compressed domain. We present a scheme for robust watermarking of MPEG-2 encoded video. The scheme is of much lower complexity than a complete decoding process followed by watermarking in the pixel domain and re-encoding. Although an existing MPEG-2 bitstream is partly altered, the scheme avoids drift by adding a drift compensation signal. The scheme has been implemented and the results confirm that a robust watermark can be embedded into MPEG encoded video which can be used to securely transmit arbitrary binary information at a data rate of several bytes/second. The scheme is easily applicable to other video coding schemes like MPEG-1, H.261, and H.263.
The author proposes an independent and novel approach to image coding, based on a fractal theory of iterated transformations. The main characteristics of this approach are that i) it relies on the assumption that image redundancy can be efficiently exploited through self-transformability on a block-wise basis, and ii) it approximates an original image by a fractal image. The author refers to the approach as fractal block coding. The coding-decoding system is based on the construction, for an original image to encode, of a specific image transformation--a fractal code--which, when iterated on any initial image, produces a sequence of images that converges to a fractal approximation of the original. It is shown how to design such a system for the coding of monochrome digital images at rates in the range of 0.5-1.0 b/pixel. The fractal block coder has performance comparable to state-of-the-art vector quantizers.
A digital watermark is an invisible mark embedded in a digital image which may be used for a number of different purposes including image captioning and copyright protection. This paper describes how a combination of spread spectrum encoding of the embedded message and transform-based invariants can be used for digital image watermarking. In particular, it is described how a Fourier–Mellin-based approach can be used to construct watermarks which are designed to be unaffected by any combination of rotation and scale transformations. In addition, a novel method of CDMA spread spectrum encoding is introduced which allows one to embed watermark messages of arbitrary length and which need only a secret key for decoding. The paper also describes the usefulness of Reed Solomon error-correcting codes in this scheme.
Research in digital watermarking has progressed along two paths. While
new watermarking technologies are being developed, some researchers are
also investigating different ways of attacking digital watermarks.
Common attacks to watermarks usually aim to destroy the embedded
watermark or to impair its detection. In this paper we propose a
conceptually new attack for digitally watermarked images. The proposed
attack does not destroy an embedded watermark, but copies it from one
image to a different image. Although this new attack does not destroy a
watermark or impair its detection, it creates new challenges, especially
when watermarks are used for copyright protection and identification.
The process of copying the watermark requires neither algorithmic
knowledge of the watermarking technology nor the watermarking key. The
attack is based on an estimation of the embedded watermark in the
spatial domain through a filtering process. The estimate of the
watermark is then adapted and inserted into the target image. To
illustrate the performance of the proposed attack we applied it to
commercial and non-commercial watermarking schemes. The experiments
showed that the attack is very effective in copying a watermark from one
image to a different image. In addition, we have a closer look at
application dependent implications of this new attack.
Embedding information into multimedia data, also called watermarking, is
a topic that has gained increased attention recently. For practical
applications like authentication and labeling in pay-per-view television
broad-casting, watermarking of video, and especially of already encoded
video, is interesting. We present a scheme for robust watermarking of
MPEG-2 encoded video. The watermark is embedded into the MPEG-2
bitstream, and can be retrieved from the decoded video. The scheme is
robust and of much lower complexity that a complete decoding process
followed by watermarking in the pixel domain and re-encoding. Although
an existing MPEG-2 bitstream is partly altered, the scheme avoids drift
by adding a drift compensation signal. The scheme has been implemented
and the results confirm that a roust watermark can be embedded into MPEG
encoded video which can be used to securely transmit arbitrary binary
information at a data rate of several bytes/second. The scheme is also
applicable to other hybrid coding schemes like MPEG-1, H.261 and H.263.
The growth of networked multimedia system has created a need for the
copyright protection of digital images and video. Copyright protection
involves the authentication of image content and/or ownership. This can
be used to identify illegal copies of an image. One approach is to mark
an image by adding an invisible structure known as a digital watermark
to the image. Techniques of incorporating such a watermark into digital
images include spatial-domain techniques, transform-domain algorithms
and sub-band filtering approaches.
The growth of the Internet and the diffusion of multimedia applications
requires the development of techniques for embedding identification
codes into images, in such a way that it can be granted their
authenticity and/or protected the copyright. In this paper a novel
system for image watermarking, which exploits the similarity exhibited
by the Digital Wavelet Transform with respect to the models of the Human
Visual System, for robustly hiding watermarks is presented. In
particular, a model for estimating the sensitivity of the eye to noise,
previously proposed for compression applications, is used to adapt the
watermark strength to the local content of the image. Experimental
results are shown supporting the validity of the approach.
Digital watermark is used to protect digital image against any illegal reproduction and tampering. In the selective block assignment process, the image is divided into N X N pixel blocks and each block is Discrete Cosine Transformed (DCT). The set of blocks will be then selectively chosen to encode the copyright message. Each selective block will be incremented by a value, in order to maintain the invisibility of the watermarking image, the incremented value should be within a range. The selection of blocks is based on measurement of the content. Depends on the amount of messages stored and the signal to noise ratio (SNR) of the resultant image required, a threshold is decided. In practice, the threshold will be set such that the duplicated message or an error correction mechanism can also be included in order to increase its robustness. The decoding process should be carried out by using the threshold values to get back the locations that have watermark information. Then the watermarked image is subtracted from the original image to obtain the secret data. Simulation results show that the watermarked image looks visually identical to the original and with an SNR of 44.7 dB for Lenna and with SNR 43 dB for airplane with size 256 X 256 pixels.
A methodology for comparing robustness of watermarking techniques is proposed. The techniques are first modified into a standard form to make comparison possible. The watermark strength is adjusted for each technique so that a certain perceptual measure of image distortion based on spatial masking is below a predetermined value. Each watermarking technique is further modified into two versions for embedding watermarks consisting of one and 60-bits, respectively. Finally, each detection algorithm is adjusted so that the probability of false detections is below a specified threshold. A family of typical image distortions is selected and parametrized by a distortion parameter. For the one-bit watermark, the robustness with respect to each image distortion is evaluated by increasing the distortion parameter and registering at which value the watermark bit is lost. The bit error rate is used for evaluating the robustness of the 60-bit watermark. The methodology is explained with two frequency-based spread spectrum techniques. The paper is closed with an attempt to introduce a formal definition of robustness.
The paper investigates the use of image histograms as watermarks. First, the problem of exact histogram specification is addressed and a method for exact histogram specification, consistent with the human perception of brightness, is developed. Next, two watermarking techniques based on exact histogram specification are proposed. The first one directly considers image histograms as watermarks. Thus, a particular histogram is assigned as a watermark and images are further transformed to have exactly the assigned histogram. Since quite large variations in image histogram are not perceived by humans, an unlimited number of invisible watermarks can be defined for which images appear visually non-distorted. Besides, by selecting histograms which are variations of uniform histogram, the transformed images are not only uniquely marked but also enhanced. The second approach conserves, for each image, its original histogram. The watermarking procedure consists of two histogram specification transforms: a transform to the assigned watermark followed by an inverse transform to recover the original histogram. Since image recovery after a histogram specification transform is not exact, the error obtained after the two consecutive transforms is further used to track each watermark.
This paper introduces the concept of Smart Images and explains the use of watermarking technology in their implementation. A Smart Image is a digital or physical image that contains a digital watermark, which leads to further information about the image content via the Internet, communicates ownership rights and the procedure for obtaining usage rights, facilitates commerce, or instructs and controls other computer software or hardware. Thus, Smart Images, empowered by digital watermarking technology, act as active agents or catalysts which gracefully bridge both traditional and modern electronic commerce. This paper presents the use of Digimarc Corporation's watermarking technology to implement Smart Images. The paper presents an application that demonstrates how Smart Images facilitate both traditional and electronic commerce. The paper also analyzes the technological challenges to be faced for ubiquitous use of Smart Images.
In our earlier work we have proposed a watermarking algorithm for JPEG/MPEG streams that is based on selectively discarding high frequency DCT coefficients. Like any watermarking algorithm, the performance of our method must be evaluated by the robustness of the watermark, the size of the watermark, and the visual degradation the watermark introduces. These performance factors are controlled by three parameters, namely the maximal coarseness of the quantizer used in re-encoding, the number of DCT blocks used to embed a single watermark bit, and the lowest DCT coefficient that we permit to be discarded. It is possible to determine these parameters experimentally. In this paper, however, we follow a more rigorous approach and develop a statistical model for the watermarking algorithm. Using this model we derive the probability that a label bit cannot be embedded. The resulting model can be used, for instance, for maximizing the robustness against re-encoding and for developing adequate error correcting codes for the label bit string.
Audio and video watermarking enable the copyright protection with owner or customer authentication and the detection of media manipulations. The available watermarking technology concentrates on single media like audio or video. But the typical multimedia stream consists of both video and audio data. Our goal is to provide a solution with robust and fragile aspects to guarantee authentication and integrity by using watermarks in combination with content information. We show two solutions for the protection of audio and video data with a combined robust and fragile watermarking approach. The first solution is to insert a time code into the data: We embed a signal as a watermark to detect gaps or changes in the flow of time. The basic idea uses numbers increasing by one. If in the verification process the next number is smaller than the last one or the step is greater than one, the time flow has been changed. This is realized without the combination of video and audio data. But we can synchronize the two data streams. A time signal is only valid if the combination of audio and video signals satisfy a certain attribute. To keep the basic example: if we embed an increasing a number in the audio and a decreasing number in the video, we could test if the combination of both always equals zero. The second solution is more complex: We use watermarks to embed information in each media about the content of the other media. With the help of speech recognition technology it is possible to embed the spoken text, the content, of an audio file in the video. With an algorithm previously developed in [1] we extract video content representation which is embedded in the audio stream. In our paper we present the problem of copyright protection and integrity checks for combined video and audio data. We show our two solutions and discuss our results.
The paper investigates the use of image histograms as watermarks. First, the problem of exact histogram specification is addressed and a method for exact histogram specification, consistent with the human perception of brightness, is developed. Next, two watermarking techniques based on exact histogram specification are proposed. The first one directly considers image histograms as watermarks. Thus, a particular histogram is assigned as a watermark and images are further transformed to have exactly the assigned histogram. Since quite large variations in image histogram are not perceived by humans, an unlimited number of invisible watermarks can be defined for which images appear visually non-distorted. Besides, by selecting histograms which are variations of uniform histogram, the transformed images are not only uniquely marked but also enhanced. The second approach conserves, for each image, its original histogram. The watermarking procedure consists of two histogram specification transforms: a transform to the assigned watermark followed by an inverse transform to recover the original histogram. Since image recovery after a histogram specification transform is not exact, the error obtained after the two consecutive transforms is further used to track each watermark.© (1999) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
In the European project SMASH a mass multimedia storage device for home usage is being developed. The success of such a storage system depends not only on technical advances, but also on the existence of an adequate copy protection method. Copy protection for visual data requires fast and robust labeling techniques. In this paper, two new labeling techniques are proposed. The first method extends an existing spatial labeling technique. This technique divides the image into blocks and searches an optimal label- embedding level for each block instead of using a fixed embedding-level for the complete image. The embedding-level for each block is dependent on a lower quality JPEG compressed version of the labeled block. The second method removes high frequency DCT-coefficients in some areas to embed a label. A JPEG quality factor and the local image structure determine how many coefficients are discarded during the labeling process. Using both methods a perceptually invisible label of a few hundred bits was embedded in a set of true color images. The label added by the spatial method is very robust against JPEG compression. However, this method is not suitable for real-time applications. Although the second DCT-based method is slightly less resistant to JPEG compression, it is more resistant to line-shifting and cropping than the first one and is suitable for real-time labeling.© (1997) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
Data hiding is the process of embedding data into image and audio signals. The process is constrained by the quantity of data, the need for invariance of the data under conditions where the `host' signal is subject to distortions, e.g., compression, and the degree to which the data must be immune to interception, modification, or removal. We explore both traditional and novel techniques for addressing the data hiding process and evaluate these techniques in light of three applications: copyright protecting, tamper-proofing, and augmentation data embedding.© (1995) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
Over the past two years, I have written seven “Legally Speaking” columns and one feature article for Communications about legal issues affecting computing professionals. These writings have covered an array of legal topics including: criminal and civil liability for hackers who breach computer security systems; first amendment issues arising in computing or electronic publishing markets; intellectual property issues, such as patent protection for computer program algorithms; copyright protection for look and feel of user interfaces; what the user interface design field thinks about such protection, and various theories by which a firm might claim to own interface specification information for software systems.
A variety of emerging applications require the design of systems for embedding one signal within another signal. We describe a new class of embedding methods called quantization index modulation (QIM) and develop a realization termed coded dither modulation in which the embedded information modulates the dither signal of a dithered quantizer. We also develop a framework in which one can analyze the performance trade-offs among robustness, distortion, and embedding rate, and we show that QIM systems have considerable performance advantages over previously proposed spread-spectrum and low-bit modulation systems.
A watermark is an invisible mark placed on an image that can be
detected when the image is compared with the original. This mark is
designed to identify both the source of an image as well as its intended
recipient. The mark should be tolerant to reasonable quality lossy
compression of the image using transform coding or vector quantization.
Standard image processing operations such as low pass filtering,
cropping, translation and rescaling should not remove the mark. Spread
spectrum communication techniques and matrix transformations can be used
together to design watermarks that are robust to tampering and are
visually imperceptible. This paper discusses techniques for embedding
such marks in grey scale digital images. It also proposes a novel phase
based method of conveying the watermark information. In addition, the
use of optimal detectors for watermark identification is also proposed
MPEG (ISO Moving Picture Expert Group) is a compression standard for video processing and is widely used in multimedia application, e.g. VOD (video on demand), HDTV (High Definition Television), and DVD (Digital Video Disk). Confidentiality [MS95, AG96] and Copyright protection [HG96] are two security-related issues that have recently brought many attentions, especially when the MPEG streams are transmitted over the public Internet. In this paper, we present a research prototype which integrates selective encryption schemes with watermarking techniques in one system. In particular, we propose a simple and efficient scheme to securely distribute watermarked MPEG video streams over multicast/broadcast channels. We have experimented 7 different selective encryption schemes and one selective watermarking scheme on three well-known MPEG streams: Bus, Flower, and Miss America. Our result indicates that, for applications like video-on-demand or pay-per-view, we can selectively encrypt (< 1%) of the total MPEG stream, and (<3.7%) for each individual frame in the worst case.
This paper presents a video watermarking technology for broadcast monitoring. The technology has been developed at the Philips Research Laboratories in Eindhoven in the context of the European ESPRIT project VIVA Visual Identity V eriication Auditor. The aim of the VIVA project is to investigate and demonstrate a professional broadcast surveillance system. The key technology in the VIVA project is a new video watermarking technique by the name of JAWS Just Another Watermarking System. The JAWS system has been developed such that the embedded watermarks i are invisible, ii are robust with respect to all common processing steps in the broadcast transmission chain, iii have a v ery low probability of false alarms, iv have a large payload at high rate, and v allow for a low complexity and a real-time detection. In this paper we present the basic ingredients of the JAWS technology. We also brieey discuss the performance of JAWS with respect to the requirements of broadcast monitoring.
We describe a digital watermarking method for use in audio, image, video and multimedia data. We argue that a watermark must be placed in perceptually significant components of a signal if it is to be robust to common signal distortions and malicious attack. However, it is well known that modification of these components can lead to perceptual degradation of the signal. To avoid this, we propose to insert a watermark into the spectral components of the data using techniques analogous to spread sprectrum communications, hiding a narrow band signal in a wideband channel that is the data. The watermark is difficult for an attacker to remove, even when several individuals conspire together with independently watermarked copies of the data. It is also robust to common signal and geometric distortions such as digital-to-analog and analog-to-digital conversion, resampling, and requantization, including dithering and recompression and rotation, translation, cropping and scaling. The same digital watermarking algorithm can be applied to all three media under consideration with only minor modifications, making it especially appropriate for multimedia products. Retrieval of the watermark unambiguously identifies the owner, and the watermark can be constructed to make counterfeiting almost impossible. Experimental results are presented to support these claims.
Scitation is the online home of leading journals and conference proceedings from AIP Publishing and AIP Member Societies
In this paper, methods for embedding additive digital watermarks into uncompressed and compressed video sequences are presented. The basic principle borrows from spread spectrum communications. It consists of addition of an encrypted, pseudo-noise signal to the video that is invisible, statistically unobtrusive, and robust against manipulations. For practical applications, watermarking schemes operating on compressed video are desirable. A method for watermarking of MPEG-2 encoded video is presented. The scheme is a compatible extension of the scheme operating on uncompressed video. The watermark is generated exactly in the same manner as for uncompressed video, transformed using the discrete cosine transform (DCT) and embedded into the MPEG-2 bit-stream without increasing the bit-rate. The watermark can be retrieved from the decoded video and without knowledge of the original, unwatermarked video. Although an existing MPEG-2 bit-stream is partly altered, the scheme avoids visible artifacts by addition of a drift compensation signal. The proposed method is robust and of much lower complexity than a complete decoding process followed by watermarking in the pixel domain and re-encoding. Fast implementations exist which have a complexity comparable to a video decoder. Experimental results are given. The scheme is also applicable to other hybrid transform coding schemes like MPEG-1, MPEG-4, H.261, and H.263.
Digital mass recording devices for video data will enter the consumer market soon. Service providers are reluctant to offer services in digital form because of their fears for unrestricted duplication and dissemination. Therefore adequate copy protection systems should be developed, most likely based on labeling techniques. In this paper we propose two different techniques for the real-time labeling of digital video. Both methods embed the label information directly into an MPEG compressed video bitstream. The first method embeds the label by changing variable length codes in the bitstream. The second method discards some of the high frequency DCT-coefficients of the bitstream to embed the label. The first technique is computationally less expensive than the second one, however, the second one is more robust against attacks to remove the label. We show that the label can still be extracted after MPEG re-encoding at a lower bit-rate.
In this paper we address the issue of fraud detection in multimedia distribution. We use the self embedding approach to solve the problem. Self embedding serves to detect and recover possible tampered multimedia data. To use this approach, we design a powerful high capacity data embedding algorithm with l/167(1 bit out of 167 raw image bits)-l/84 hiding ratios subject to JPEG compression with the quality factor equals to 75. With such high embedding capacity, one can easily embed some important regions of an image unto the image self. We will demonstrate how the tampered data are recovered by using this technique. in an image to counterattack possible tampering. Fridrich’s self embedding method embeds some important regions of an image into the Least Significant Bits (LSB) of the DCT coefficients of the image. The embedding process modifies the DCT coefficients in an ad hoc way. They are very sensitive to any attacks and image processing algorithms including JPEG compression. Here we will develop a general high capacity data embedding algorithm that is robust to some moderate attacks, and can be used as an self embedding tool. The highlights of this work are: 1. A fast and efficient high capacity data embedding scheme. Our method provides the best compromise of the robustness, data capacity, and visual quality. 2. F’raud detection and tampered data recovering. Our method provides the solution to detect any alleged tampered images, and recover the original data as well.
The use of digital data has become more and more commercialized. This is especially true for digital images, where proofs of origin and of content integrity are an important issue. This paper describes a problem related to 'proof of origin' and proposes a possible solution to it. After a discussion of the solution, possible extensions and related areas of work are addressed.
Digital watermarks have been proposed as a method for discouraging
illicit copying and distribution of copyright material. This paper
presents a new approach for the secure and robust copyright protection
of digital images. The digital watermarks described in this paper are
designed to be, as far as possible, invariant against image
transformations such as rotation, translation, scaling and cropping. We
concentrate especially on the desirable properties of the Fourier
Transform and propose a novel technique based on an invisible template
which allows us to reverse many of the effects of image processing on
the digital watermark. Robustness of the watermark to operations such as
lossy compression is achieved by using a perceptually adaptive spread
spectrum communications approach, in which a spread spectrum signal is
embedded in selected components of the magnitude spectrum of the image.
The keys used to embed the spread spectrum signal are generated,
certified, authenticated and securely distributed using a public key
infrastructure containing an electronic copyright office and a
certification authority. The security architecture used for this purpose
is also outlined.